COMPARATIVE EVALUATION OF CHEMICAL RANKING …isse.utk.edu/ccp/pubs/pdfs/CECRSM.pdf · OF CHEMICAL RANKING AND SCORING METHODOLOGIES ... octanol-water partition coefficient ... Chemicals

COMPARATIVE EVALUATIONOF CHEMICAL RANKING

AND SCORING METHODOLOGIES

EPA Order No. 3N-3545-NAEX

Prepared by

University of TennesseeCenter for Clean Products and Clean Technologies

Gary A. Davis, Principal InvestigatorMary Swanson, Senior Research Associate

Sheila Jones, Research Assistant

April 7, 1994

Printed on Recycled Paper

ii

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

LIST OF ACRONYMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

SECTION 1: INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1SECTION 2: METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1 SUMMARY OF SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.2 DETAILED DESCRIPTION OF SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.3 DESCRIPTION OF IMPORTANT ELEMENTS AND APPLICATIONS

OF RANKING SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.3.1 Purpose and Application of Chemical Ranking and Scoring Systems . . . . . 102.3.2 Human Health Criteria and Endpoints . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.3.3 Criteria and Endpoints for Environmental Effects . . . . . . . . . . . . . . . . . . . . 222.3.4 Measures of Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242.3.5 Missing Data, Data Selection Approach . . . . . . . . . . . . . . . . . . . . . . . . . . 292.3.6 Aggregation and Weighing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312.3.7 Other Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

SECTION 3: DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.1 RISK ASSESSMENT PRINCIPLES APPLIED TO CHEMICAL

RANKING AND SCORING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383.2 SIMILARITIES AND DIFFERENCES AMONG SYSTEMS REVIEWED . . . . 39

3.2.1 Most Common Effects and Exposure Endpoints Used . . . . . . . . . . . . . . . 393.2.2 Ranking versus Categorization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.2.3 Quantitative versus Qualitative Endpoints . . . . . . . . . . . . . . . . . . . . . . . . . 413.2.4 Assigning Scores to Endpoint Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.3 GENERAL STRENGTHS AND WEAKNESSES OF EVALUATEDSYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

SECTION 4: CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

APPENDIX A SUMMARIES OF RANKING AND SCORING SYSTEMS . . . . . . . . A-1

APPENDIX B RANKING AND SCORING SYSTEMS IDENTIFIED . . . . . . . . . . . . . . . . . B-1

APPENDIX C DETAILED TABLES OF CRITERIA, ENDPOINTS,AND DATAAPPROACH USED BY EACH EVALUATED SYSTEM . . . . . . . . . . C-1

iii

LIST OF TABLES

Table No. page

1. Chemical Ranking and Scoring Systems Evaluated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42. Endpoints Used for Scoring Environmental and Human Health Effects . . . . . . . . . . . . . . . . . 163. Endpoints Used for Scoring Exposure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264. WP Factors (WP) Scores for Several Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

B-1 Overview of Purpose and Developers/Users of Ranking and ScoringSystems Evaluated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-1

B-2 Overview of Purpose and Developers/Users of Additional Rankingand Scoring Systems Identified . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B-2

C-1 Toxicity Criteria and Endpoints Used in Evaluated Systems . . . . . . . . . . . . . . . . . . . . . . . C-1C-2 Overview of Exposure Criteria and Endpoints Used in Evaluated Systems . . . . . . . . . . . C-26C-3 Overview of Data Selection Approach Used in Evaluated Systems . . . . . . . . . . . . . . . . C-33

LIST OF FIGURES

Figure No. page

1. Comparison of Scoring for Several Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 432. Development of a Chemical Ranking and Scoring System . . . . . . . . . . . . . . . . . . . . . . . . . . 47

iv

LIST OF ACRONYMS

AALAC: ambient aquatic life advisory concentrationARET: Accelerated Reduction/Elimination of Toxics ADI: allowable daily intakeATSDR: Agency for Toxic Substances and Disease RegistryAWQC: aquatic water quality criteriaBAF: bioaccumulation factorBCF: bioconcentration factorBP: boiling pointBUA: Beratergremium für Umweltrelevante AltstoffeCAG: carcinogen assessment groupCCHTE: chemical category human toxicity estimateCFC: chlorofluorocarbonCMR: Critical Materials RegisterDfE: Design for the EnvironmentEC: European CommunitiesEC50: the median effect concentration; the concentration at which 50% of the test population exhibit aspecified response during a specified time periodED10: estimated dose associated with a lifetime increased cancer risk of 10%EEC: European Community CommissionELU: Environmental load unitsEMPPL: Effluent Monitoring Priority Pollutants ListEPA: United States Environmental Protection AgencyEPS: environmental priority strategiesERRP: ecological risk reduction potentialGWU: George Washington UniversityHA: health advisoryHEAST: Health Effects Assessment Summary TablesHHSATR: human health structure activity team rankHPV: high production volumeHRRP: human risk reduction potentialHRS: Hazard Ranking SystemHWQC: human health water quality criteriaIARC: International Agency for Research on CancerIC50: inhibitory concentration for 50% test populationIRIS: Integrated Risk Information SystemITC: Interagency Testing CommitteeKow: octanol-water partition coefficientLC50: median lethal concentration; the concentration at which 50% of the test population die during aspecified time periodLCA: life cycle assessmentLD50: lethal dose for 50% test populationLDlo: lethal dose to some value < 50%

v

LIST OF ACRONYMS, continued

LET: lethalityLOAEL: lowest observable adverse effect levelMATC: maximum acceptable toxicant concentrationMED: minimum effective doseMOE: Ministry of the Environment (Ontario)NA: not applicableNAAQS: national ambient air quality standardNEL: no effect levelNFPA: National Fire Protection AssociationNOAEL: no observable adverse effect levelNOEC: no observable effect concentrationNOEL: no observable effect levelNOx: nitrous oxidesNPL: National Priority ListNTP: National Toxicology ProgramOECD: Organization for Economic Cooperation and DevelopmentPGR: population growth rateq1*: cancer potency slope factor QSAR: quantitative structure-activity relationshipRfC: reference concentrationRfD: reference doseRQ: reportable quantitySAR: structure-activity relationshipSARA: Superfund Amendments and Reauthorization ActSATR: structure activity team rankSIDS: screening information data setSRS: Site Ranking SystemTDlo: total dose resulting in a sublethal effectTPQ: threshold planning quantityTRI: Toxics Release InventoryTSCA: Toxic Substances Control ActTSCA CSSC: TSCA chemical scoring system categoryUCR: unit cancer riskURF: unit risk factorUT: University of Tennessee (Knoxville, Tennessee)VOC: volatile organic compoundWMS: Wet Milieugevaarlijke StoffenWOE: weight of evidence

1 "Data selection approach" is a term used here to describe the methods used to build a comparabledata set for a specific measurable characteristic across a set of chemicals of interest. As an example, tocharacterize the bioconcentration potential for a set of 100 organic compounds, measuredbioconcentration factors are selected first for all chemicals where this information is available. For allother chemicals, a bioconcentration value is estimated from the octanol-water partitioning coefficient orfrom water solubility data.

1

SECTION 1: INTRODUCTION

Between 60,000 and 100,000 of the over 8,000,000 chemicals listed by the Chemical AbstractsServices Registry are commercially produced and are potential environmental pollutants. Some kind ofrisk-based evaluation for these chemicals is often required to evaluate the impacts of chemical use orreleases, for regulatory action, and to set priorities for pollution prevention. The time and resources,however, are not reasonably available to test all of these chemicals for their potential health andenvironmental effects. During the last decade there have been vast improvements in the methods usedto test chemicals for toxicity and environmental fate and to interpret these data within a risk assessmentframework. We have not, however, developed generally accepted and widely used tools to betterenable us to set priorities and focus limited resources on selecting those chemicals for study that wouldyield the greatest environmental benefits.

Risk-based ranking and scoring systems can be used to focus attention and resources on the largestpotential hazards. Risk-based chemical ranking and scoring combines an assessment of both the toxiceffects of chemicals (human and/or environmental) and the potential exposure to those chemicals toprovide a relative evaluation of risk. Along with toxicity and exposure, ranking and scoring systemsmay include other environmental impacts (e.g. ozone depletion) and some measure of economic impactand/or societal value.

Although numerous ranking and scoring systems have been or are being developed, there is currentlyno scientific consensus on risk ranking methods. Chemical risk ranking has received the most attention,and several systems have been used, for example, to determine which chemicals should be included invarious regulatory pollutant lists. To facilitate development of a framework for overall human health andenvironmental risk ranking, this report presents an evaluation of existing chemical ranking and scoringsystems. Approximately fifty systems are compared and evaluated in terms of:

• the purpose and application of the ranking and scoring system;• the human health criteria and endpoints included;• the criteria and endpoints included for environmental effects;• whether measures of exposure are included;• the data selection approach1 and handling of missing data;

2

• the use of aggregation and weighing of different health and environmental impacts;• methods of accounting for chemical potency and severity of effects; and• inclusion of other impacts or issues.

An evaluation of the important issues inherent to the development of any consensus ranking and scoringsystem is presented, and important similarities and differences among the systems evaluated arediscussed.

In this document, following the introduction, Section 2 presents the methods used by the evaluatedsystems. This includes a list and summary of the systems, a more detailed description of the systems,and a description of important elements and applications of ranking systems. Section 3 contains adiscussion, based on the analysis of systems in Section 2, of the application of risk assessmentprinciples to chemical ranking and scoring, similarities and differences among the systems, and generalstrengths and weaknesses of the evaluated systems. Section 4 presents conclusions and suggestions forfuture work.

3

SECTION 2: METHODS

2.1 SUMMARY OF SYSTEMS

Fifty-one ranking and scoring systems have been reviewed for this report (listed in Table 1). Asummary is provided for each of these in Appendix A. The systems selected for evaluation span a widerange of methodologies and levels of complexity (from simple and straight-forward to fantasticallyelaborate). Evaluating this variety of systems is intended to provide a representative sampling of theapproaches to chemical ranking and scoring and the issues involved in developing a standardized (orconsensus) framework.

It is necessary to distinguish chemical ranking and scoring systems from other risk ranking systemssuch as those that primarily rank sites (e.g. potential sites for listing on the National Priorities List) orissues (e.g. risk from hazardous waste sites versus air pollution, etc.). The majority of ranking andscoring systems that we have evaluated are based on a chemical-specific approach, which involvesranking, scoring, or categorizing a list of chemicals based on chemical-specific attributes. A system isconsidered a chemical ranking or scoring system if it meets the following criteria:

• it ranks or scores a list of chemicals;• it results in a relative ranking or scoring, not quantitative measure(s) of risk;• it includes measures of either toxicity alone or toxicity and exposure; or• if it is a subsystem of a site or issue ranking method, the approach ranks sites or issues

based on an initial chemical-specific evaluation.

Thirteen of the fifty-one systems reviewed here did not meet these conditions. The following systemsreviewed did not meet the criteria of a chemical scoring or ranking system as defined above:

• CERCLA Hazard Ranking System (HRS) (EPA 1990): a site ranking system;• Modified Hazard Ranking System (Hawley & Napier, 1985): a site ranking system;• A Groundwater Pollution Priority System (Hutchinson & Hoffman): a site ranking

system;• EPA Unfinished Business Report (Morgenstern et. al., 1987): used to rank

environmental issues which pose societal risks;• Coastal Hazardous Waste Site Review (Beckvar & Harris, 1985): a site ranking

system;• Site Ranking System (SRS) for Chemical and Radioactive Waste (Rechard et. al.,

1988, 1991): a site ranking system;• Evaluating Contamination Potential of Surface Impoundments (Silka &

Swearingen, 1978): a site ranking system;

4

TABLE 1. List of Ranking and Scoring Systems Evaluated

Systemno. a System Name or Title Reference

1 TRI Environmental Indicators Methodology (draft) Abt Associates, Inc., 1992

2 ATSDR, "CERCLA Section 104 Third Priority List" ATSDR, 1992

3 Existing Chemicals of Environmental Relevance Behret, 1989a

4 Existing Chemicals of Environmental Relevance II, Selection Criteriaand Second Priority List

Behret, 1989b

5 Review of Region VII TRI Strategy Bouchard, 1991

6 Candidate Substance List for Bans or Phase-outs Socha et al, 1992

7 Criteria Identifying High Risk Pollutants BNA, 1991

8 A Classification System for Hazardous Chemical Wastes Crutcher & Parker, 1990

9 CERCLA Hazard Ranking System (HRS) EPA, 1990

10 Identifying Chemical Candidates for Sunsetting: George WashingtonUniversity

Foran & Glenn, 1993

11 Existing Chemicals: Systematic Data Collection and Handling forPriority Setting

Gjøs et al., 1989

12 Substances and Preparations Dangerous for the Environment: ASystem for Classification, Labeling and Safety Data Sheets

Gustafsson & Ljung, 1990

13 Notes on Ranking Chemicals for Environmental Hazard Halfon & Reggiani, 1986

14 Application of the Hazard Ranking System to the Prioritization ofOrganic Compounds Identified at Hazardous Waste Remedial ActionSites

Hallstedt et al., 1986

15 Modified Hazard Ranking System (mHRS); A Ranking System forHazardous Waste Sites with Mixed Radioactive and HazardousWastes

Hawley & Napier, 1985

16 A Groundwater Pollution Priority System (GWPPS) Hutchinson & Hoffman,1983

17 The Great Lakes Water Quality Agreement (GLWQA) Annex 1, Lists1,2,3

IJC, 1989



5

18 Chemical Scoring by a Rapid Screen of Hazard (RASH) Method Jones et al., 1987

19 Systematic Approach for Environmental Hazard Ranking of NewChemicals ("Schmallenberg")

Klein et al., 1988

20 WMS Scoring System Könemann & Visser, 1988;Timmer et al., 1988

21 Benchmark Comparisons Laskowski et al., 1982

22 Michigan Critical Materials Register (MCMR) Michigan DNR, 1987

23 USEPA Unfinished Business Report: A Comparative Assessment ofEnvironmental Problems

Morgenstern et al., 1987

24 Chemical Scoring System for Hazard and Exposure Identification O'Bryan & Ross, 1988

25 Effluent Monitoring Priority Pollutant List (EMPPL) Environment Ontario, 1987,1988

26 Coastal Hazardous Waste Site Review Beckvar & Harris, 1985

27 Site Ranking System (SRS) for Chemical and Radioactive Waste Rechard et al., 1988;Rechard et al., 1991

28 A Practical Method for Priority Selections and Risk AssessmentAmong Existing Chemicals

Sampaolo & Binetti, 1986,1989

29 UT Chemical Ranking System (draft) Davis et al., 1993

30 A Manual for Evaluating Contamination Potential of SurfaceImpoundments

Silka & Swearingen, 1978

31 The EPS Enviro-Accounting Method Steen & Ryding, 1992

32 Defense Priority Model (FY 1993 Version) U.S. DOD, 1991

33 Hazardous Air Pollutants: Proposed Regulations GoverningConstructed, Reconstructed and Modified Major Sources

EPA, 1993b (40 CFR 63)

34 Ranking System for Clean Water Act Section 307(a) List of PriorityPollutants

Poston & Prohammer, 1985;Cornaby et al, 1986



6

35 CERCLA Section 102 Reportable Quantity Ranking Process EMS, 1985;EPA, 1989c

36 The Source Category Ranking System Radian Corp., 1990

37 Sax Toxicity Ratings (or Hazard Index) Sax & Lewis, 1989

38 Examination of the Severity of Toxic Effects and Recommendationsof a Systematic Approach to Rank Adverse Effects

Environ 1986

39 Screening Procedure for Chemicals of Importance to the Office ofWater

EPA, 1986

40 Measuring Air Quality: The New Pollutants Standards Index EPA, 1978

41 Ranking the Relative Hazards of Industrial Discharges to POTWsand Surface Waters

Abt Assoc., 1991

42 Targeting Pollution Prevention Opportunities Using the 1988 TRI ICF, 1990

43 EPA Design for the Environment Program Use Cluster ScoringSystem (also called: Chemical Use Clusters Scoring Methodology)

EPA, 1993a

44 Screening Methodology for Pollution Prevention Targeting EPA (date unknown)

45 Toxic Chemical Release Inventory Risk Screening Guide EPA, 1989a

46 TSCA's TRI Chemical Risk Assessment Pre-Screening Methodology EPA (date unknown)

47 Priority Setting of Existing Chemicals ("Schmallenberg") Weiss et al., 1988

48 Multi-Media Environmental Pollutant Assessment System(MEPAS), formerly the Remedial Action Priority System (RAPS)

Whelan et al., 1987; Droppoet al., 1989; Strenge et al.,1989;Whelan et al., 1992

49 Canadian Accelerated Reduction/Elimination of Toxics (ARETS)Scoring Protocol

CLC, 1992; Canadian ARET,1993

50 Risk Assessment Guidance under CERCLA, screening chemicals ofpotential concern

EPA, 1989d



7

51 EC Proposal for Priority Setting of Existing Chemical Substances(Draft)

van der Zandt & vanLeeuwen, 1992

(a) System numbers are used as a shorthand method of tracking the systems. They were based on a preliminaryalphabetical list of references, but at this time are essentially an arbitrarily assigned number.

8

• The EPS Enviro-Accounting Method (Steen & Ryding, 1992): developed for lifecycle impact assessment;

• Defense Priority Model (U.S. DOD, 1991): a site ranking system;• Examination of the Severity of Toxic Effects (Environ Corp., 1986): not chemical

specific;• Ranking the Relative Hazards of Industrial Discharges to POTWs and Surface

Waters (Abt Assoc., 1991): a site ranking system;• Targeting Pollution Prevention Opportunities Using the 1988 TRI (ICF, 1990): a

site ranking system; and• Multi-media Environmental Pollutant Assessment System (MEPAS) (Whelan et.

al., 1987; Droppo et. al., 1989; Strenge et. al., 1989; Whelan et. al., 1992): a siteranking system.

For the sake of completeness, and to illustrate the diversity among what are called risk ranking systems,these thirteen systems are also summarized in Appendix A.

2.2 DETAILED DESCRIPTION OF SYSTEMS

Descriptions of the fifty-one systems have been distilled into two-page summaries (Appendix A) thatinclude the following information:

• system name or title;• publication references;• the intended user of the system;• the purpose or application;• chemicals addressed or used for demonstration;• summary of method or algorithm;• an outline of the criteria, subcriteria and endpoints used; and• data selection approach.

Several other surveys of existing (primarily chemical) ranking and scoring systems have been identified:

• Waters et. al. (1993) describes seventeen ranking and scoring systems, including ninemethods for ranking existing sites and eight "various chemical and substance rankingmethodologies";

• Foran and Glenn (1993) analyze eight "existing chemical scoring systems" in detail, andinclude summary tables of effects criteria, endpoints, and scoring methods for thescreening systems analyzed;

9

• ICF (1993) provides detailed summaries and comparisons of five chemical scoringsystems with brief descriptions of fifteen additional systems in a bibliography;

• Abt Associates, Inc. (1992) summarizes, in some detail, twenty-one U.S.Environmental Protection Agency (EPA) ranking systems. An additional thirteen"ranking and indexing efforts" from outside of the EPA are described briefly. All of thesystems surveyed make use of TRI data;

• The Organization for Economic Cooperation and Development (OECD, 1986)describes fifteen ranking and scoring systems in some detail and lists an additional fifty"known priority ranking systems" by system name, developer, and intended user of thesystem. Full references for the fifty systems were not included;

• Environ Corp. (1986) describes thirty-four "chemical ranking schemes", includingeleven from the EPA, eight from other federal agencies, four state ranking schemes, two"degree of hazard" ranking schemes, three from international agencies, two "otherranking schemes", and four carcinogen ranking schemes; and

• Hushon and Kornreich (1984) summarize thirty-four "existing chemical scoringsystems" in a table that includes the system title, purpose, developer(s), user(s), criteriaused for scoring, algorithms used, substances scored, and references.

When all of the systems described by these authors are included, a total of 147 ranking and scoringsystems have been identified. A complete listing of these is provided in Appendix B, with the systemname, the developers and intended users of the system, the purpose or application of the system, andthe system approach (where the information was available).

2.3 DESCRIPTION OF IMPORTANT ELEMENTS AND APPLICATIONS OFRANKING SYSTEMS

Important components of any chemical ranking and scoring system include:

• the purpose and application of the ranking and scoring system;• the human health criteria and endpoints included;• the criteria and endpoints included for environmental effects;• whether measures of exposure are included;• the data selection approach and handling of missing data; and• the use of aggregation and weighing of different impacts.

10

Other issues addressed include major ecological impacts, chemical mixtures, application to life cycleassessment (LCA), and economic and social considerations. These are discussed below with examplesof various approaches used by the evaluated systems.

2.3.1 Purpose and Application of Chemical Ranking and Scoring SystemsChemical ranking and scoring systems are typically used as screening tools for a rapid assessment ofrelative chemical hazards. They consider the toxic effects of chemicals and some measure of exposure,but are not intended to serve the same purpose as a quantitative risk assessment.

The purpose or intended use of the ranking and scoring system affects all of the aspects of the system,including the human health and environmental impacts considered, the use of aggregation and weighing,and how missing data are handled. Chemical ranking and scoring systems have been developed for thebroad purposes of regulatory action, priority setting, and impact evaluations. Each of these is discussedbelow.

Regulatory ActionThe need for the regulation of chemicals of concern has been the driving force behind the developmentof many of the chemical ranking and scoring systems evaluated. Examples of this application aredescribed below:

The CERCLA Section 102 Reportable Quantity Ranking Process (EMS, 1985; EPA, 1989)established reportable quantities (RQs) through a scoring process that takes into account environmentaland human health hazards of chemicals on the CERCLA list. RQs are threshold quantities at whichcertain chemical releases must be reported to the EPA.

The Michigan Critical Materials Register (CMR) (MDNR, 1987) ranking process results in a list ofchemicals that may threaten water quality in Michigan. Chemicals included in the register areconsidered to pose a high degree of environmental concern, and companies must report their use anddischarge of these chemicals.

The Ontario Ministry of the Environment (MOE) Candidate Substance List for Bans or Phase-outs (Socha et. al., 1992) is a ranking system that identifies chemical substances for release reduction,bans, or phase-outs.

The George Washington University (GWU) scoring system (Foran and Glenn, 1993) is intended toserve as a tool for pollution prevention in the Great Lakes region through the identification of chemicalsubstances for Sunsetting (bans or phase-outs) and other activities.

11

Priority SettingSeveral ranking and scoring systems have been developed specifically for assessing chemical hazardsfor priority setting for non-regulatory purposes. For example:

The EPA Design for the Environment Program (DfE) Use Cluster Scoring System (EPA, 1993) isaimed at measuring potential health and environmental risks of chemicals that are used in industryclusters (i.e., certain common industrial processes) as a way of evaluating the benefits of substitutes forthose chemicals. The relative risk scores are provided to industry as a means of encouraging thevoluntary adoption of pollution prevention measures.

The Wet Milieugevaarlijke Stoffen (WMS) Scoring System (Könemann and Visser, 1988; Timmeret. al., 1988) was developed by the Directorate General for Environmental Protection of theNetherlands Ministry of Housing, Physical Planning, and the Environment. It was intended for use inindustry, government, and academia for the selection of a limited number of chemicals as priorities forfurther investigation.

The Agency for Toxic Substances and Disease Registry (ATSDR) System (ATSDR, 1992): TheSuperfund Amendments and Reauthorization Act (SARA) of 1986 required the ATSDR to prepare alist, in order of priority, of the hazardous substances commonly found at National Priority List sites thatpose the most significant potential human health threats. Substances on this priority list becomecandidates for the preparation of toxicological profile reports prepared by ATSDR.

The University of Tennessee (UT) Center for Clean Products and Clean Technologies chemicalranking system (Davis et. al., 1993) prioritizes chemicals for safe substitutes assessments. The systemhas also been used for impact evaluation, as discussed below, and for scoring of TRI releases fromchemical production facilities.

Impact EvaluationOther scoring systems have been developed to evaluate the potential impacts of chemical releases.

The Toxics Release Inventory (TRI) Environmental Indicators Methodology (Abt Assoc., 1992) isintended to evaluate TRI releases and derive a value to indicate the overall impacts of those releases byall facilities and to each environmental medium. Annual calculations of the indicator numbers allow acomparison of potential TRI impacts from year to year.

The UT ranking system (Davis et. al., 1993) has been modified to allow a comparative assessment ofthe potential hazard or risk posed by aggregate TRI chemical releases and transfers from an entire stateor from facilities (Kincaid and Bartmess, 1993). The modified system was used to assess, on a relativebasis, the potential impacts from chemical releases for the five states with greatest releases (Tennessee,Texas, Louisiana, Indiana, and Ohio) and is being used to score TRI releases from chemical productionfacilities.

12

The New Pollutant Standards Index (EPA, 1978) was developed for use by local and state airpollution control agencies to provide a simple method for reporting daily air pollutant concentrationsand to inform the public about the potential health effects associated with these concentrations. Theindex is calculated from a measured air pollutant concentration and its national ambient air qualitystandard, and is characterized with terms such as "good" or "unhealthful".

Another aspect of impact assessment is within the field of life cycle assessment (LCA). The methodsfor conducting life cycle impact assessment are currently being developed (SETAC, 1993). Chemicalranking and scoring methods with a focus on impact assessment have the potential to be applied to thisemerging technology. LCA is discussed further in section 2.3.7.

2.3.2 Human Health Criteria and EndpointsIn order to assess the potential or actual hazard associated with a particular chemical substance,chemical ranking and scoring systems may include criteria for scoring the toxicity of a chemical toterrestrial mammals, non-mammalian terrestrial species, aquatic organisms, and/or plants. The toxicityof chemicals to these organisms is used to evaluate potential effects to human health and theenvironment. Acute, subchronic and chronic mammalian toxicity data are often used as surrogates forhuman health effects. Ecological effects typically include acute, subchronic and chronic toxicity toterrestrial mammals, non-mammalian terrestrial species, aquatic organisms and plants. Although someranking and scoring systems include bioaccumulation and/or persistence in the environmental effectscategory, these are discussed within the framework of this paper as factors affecting exposure.

Overview of Human Health EffectsHuman health effects include many responses in humans caused by chemical exposure. Epidemiologicaldata may be used to characterize health effects, but laboratory mammalian toxicity data are most oftenincluded in chemical ranking and scoring systems. The numbers and types of endpoints used to assesspotential health effects vary significantly.

Criteria for evaluating health effects often include toxicity resulting from varying durations of exposure. Those effects resulting from acute, sub-chronic and chronic exposure are commonly included inchemical ranking and scoring systems. A brief description of each of these types of effects is listedbelow:

Acute Effects. Acute toxicity tests usually involve a single dose and a 14-day observation period. These tests are most commonly conducted on the mouse or rat, but other species, such as the dog orrabbit, may be used. Acute effects are often characterized by lethality, commonly reported as themammalian median lethal dose or concentration (LD50 or LC50). This is the dose or concentrationrequired to elicit lethality in 50% of the animals tested. Non-lethal acute effects are sometimes includedas well. Skin or eye irritation and sensitization are examples of such effects. Routes of administrationcommonly preferred include oral, dermal and inhalation exposure. Some systems use data from otherroutes of exposure in the absence of preferred data.

13

Sub-chronic Effects. The most common test duration is 90 days for these tests, but the exposure timemay vary. The main goal of these studies is to determine the no-observed-effect-level (NOEL) and toidentify the specific organs affected after repeated doses of the test substance. These tests are usuallyconducted on the rat or the dog by an oral route of administration, but other species and routes may beused (Klaassen and Eaton, 1993).

Chronic Effects. These are long-term (longer than 3 months) studies designed to assess the cumulativetoxicity of chemicals (Klaassen and Eaton, 1993). Chronic test data are often utilized in chemicalscoring systems. Chronic health effects may be evaluated on the basis of a wide variety of toxicologicalendpoints. Carcinogenicity, mutagenicity, teratogenicity, reproductive toxicity, and other chronic toxiceffects are often included in chemical scoring systems. General chronic effects may be characterized byRQs, reference doses (RfDs), threshold planning quantities (TPQs), the minimum effective dose(MED), or other measures.

A brief description of chronic effects often assessed in chemical ranking and scoring is provided below:

Carcinogenicity. Chemical carcinogens are substances which cause cancer in humans or other animals. Carcinogenicity is often evaluated on the basis of weight-of-evidence classifications developed by theEPA and/or strength-of-evidence classifications by the International Agency for Research on Cancer(IARC). Weight-of-evidence classification considers all long-term animal and relevant human studiesas well as metabolism, pharmacokinetic and mechanistic information, structure-activity relationships(SARs) and other studies involving biochemical or physiological function. Strength-of-evidenceclassification may refer to the magnitude of conviction about the results of an experiment. For example,the National Toxicology Program (NTP) classifies each carcinogenic bioassay according to the amountand type of data from an experiment (Scala, 1993). The strength-of-evidence scheme developed byIARC excludes mechanistic information relating to the relevance to humans of bioassay data whichshow evidence of carcinogenesis in animals. This scheme does not consider all relevant data. Forthese reasons, the IARC classification scheme is considered to be based on strength-of-evidence ratherthan weight-of-evidence (Ashby et. al., 1990).

Sometimes one or both of the EPA and IARC classifications are combined with a measure of potency,usually the slope factor (q1

*) or the ED10 in chemical ranking and scoring systems. Potency refers to therelationship between the dose and the response. The slope factor is an upper bound estimate of theprobability of a response per unit intake of a chemical over a lifetime. It is used to estimate theprobability of an individual developing cancer resulting from a lifetime exposure to a carcinogen (EPA1989b). This is the potency factor normally used in EPA risk assessments. The ED10 is the estimateddose associated with a lifetime cancer risk increase of 10 percent. This is used in establishing RQsunder CERCLA Section 102.

Mutagenic Effects. Mutagenesis occurs when chemicals cause changes in the genetic material whichcan be transmitted during cell division (Klaassen and Eaton, 1993). Several procedures, both in vitro

14

and in vivo, have been developed to test chemicals for possible mutagenicity. Tests for mutagenicityare often used to screen for potential carcinogenesis because the initiation of chemical carcinogenesis isbelieved to be a mutagenic occurrence (Klaassen and Eaton, 1993). Like carcinogenicity, mutagenicityis often evaluated according to weight-of-evidence classification which is sometimes combined with ameasure of potency and/or severity. Mutagenic effects may also be of interest in terms of geneticalterations in the next generation.

Reproductive Toxicity. This refers to the occurrence of adverse effects, resulting from exposure tochemical or physical agents, on the male or female reproductive system. These may include effects onfertility, gestation, or lactation, among others (Klaassen Eaton, 1993). These effects are usually scoredaccording to some type of weight-of-evidence and/or potency and severity.

Teratogenic Effects. Teratogenicity occurs when exposure to some chemical or physical agent inducesdefects during the development of an organism from conception to birth (Klaassen and Eaton, 1993). Teratogenicity is sometimes considered separately from other reproductive effects. Likewise, theseeffects are often scored on the basis of weight-of-evidence and/or some measure of potency and/orseverity.

Neurotoxicity. This includes adverse effects on the nervous system caused by chemical exposure whichmay be structural and\or functional and may include behavioral changes and learning disabilities.

Several chemical ranking systems include toxicity scoring criteria based not on specific toxicologicalendpoints, but on indices that combine or scale one or more toxicological endpoints, such as:

• RQ: A hazardous substance, under CERCLA, released in a quantity above this amountmust be reported to the National Response Center, the State Emergency ResponseCommission and the Local Emergency Planning Committee;

• RfD: An estimated daily exposure to the human population that is likely to be withoutappreciable risk or adverse effects in a lifetime, expressed in mg/kg/day; and

• TPQ: The amount of an extremely hazardous substance present at a facility abovewhich the emergency planning notification must be provided to the State EmergencyResponse Commission and the Local Emergency Planning Committee.

15

Endpoints Used for Human Health EffectsMember countries of the OECD have undertaken the investigation of approximately 600 highproduction volume (HPV) chemicals. A data set has been established which is necessary to provide aninitial screening of health and environmental risks. The Screening Information Data Set (SIDS) is theminimum information needed for deciding whether or not a HPV chemical should be considered of lowcurrent concern, considered for further information gathering or testing, or a candidate for furtherreview with possible action to reduce risks posed by the chemical (Auer, 1992). This represents aneffort to standardize, on an international basis, the basic experimental data needed to characterize thepotential health and environmental effects of chemicals.

The SIDS database (Auer, 1992) includes the following toxicological data:

• acute toxicity;• repeated dose toxicity;• genetic toxicity (two endpoints, generally point mutation and chromosomal aberrations);

and• reproductive toxicity (including fertility and developmental toxicity).

The endpoints used to score human health and ecological effects are presented in Appendix C and inTable 2. (The use of criteria and endpoints is discussed further in section 3.2.1.) Although mostchemical ranking and scoring systems evaluated were designed with human health protection in mind,most of them do not clearly distinguish between endpoints included to assess human health from thoseused to characterize toxicity to other terrestrial mammals. Often a score for 'general toxicity' isdetermined. The manner in which systems evaluate health effects varies dramatically among the manydifferent systems. As can be seen, the specific endpoints selected to represent health effects arenumerous.

The Nordic System (Gjøs et. al., 1989) scores health effects based on data pertaining to acutemammalian toxicity, irritation, sensitization, general toxicity, genotoxicity, carcinogenicity andreproduction damage/teratogenicity.

The EPA DfE Use Cluster Scoring System (EPA, 1993) assigns a human health hazard ranking forchemicals. This system assigns a score on the basis of criteria such as the RfD, RQ, or the TPQ,among others. The carcinogenic properties of each chemical are also scored and the higher of the twoscores is selected as the human health hazard potential score.

16

TABLE 2. Endpoints Used for Scoring Environmental and Human Health Effects

Endpoint System Number for Method Using this Endpoint a

Plant Toxicity (8 systems)b

phytotoxicity in higher plants (qualitative) 28

qualitative effects (algae) 28, 47

aquatic, terrestrial IC50 17

EC50 6, 22, 25, 47, 49

NOAEL/NOAEC 6, 25, 49

Aquatic Toxicity (26 systems)

AWQC (acute or chronic) 1, 9, 42, 45

AALAC (ambient aquatic life advisory concentration)

1, 9

NOAEL 1, 6, 10, 25, 49

acute LC50 1, 2, 3, 4, 5, 6, 9, 10, 11, 12, 17, 19, 20, 22, 24, 25, 28, 29, 34,35, 39, 49

acute LD50 17

acute EC50 3, 4, 10, 11, 12, 17, 19, 20, 22, 24, 28,

subchronic or chronic EC50 6, 25, 34, 49

subchronic or chronic MATC 6, 22, 25, 34, 49

NOEL/NOEC 20, 24, 29

aquatic toxicity RQ 41, 42

population growth rate (PGR) (algae) 5

lethality (LET) (algae) 5

acute and prolonged fish, Daphnia toxicity 47

Terrestrial (non-mammal) Toxicity (8 systems)

subchronic or chronic effects 6

acute LC50 22

acute LD50 10, 22, 28

chronic NOAEL 10

severity and effective dose 22

subchronic or chronic NOEL 25, 49

acute toxicity (bird, earthworm) 47



17

unspecified 11

Mammalian Toxicityc (19 systems)

acute LD50 2, 3, 4, 6, 7, 10, 11, 12, 17, 19, 20, 24, 25, 28, 29, 34, 35, 49

acute LC50 2, 3, 4, 6, 7, 11, 12, 17, 24, 25, 28, 35, 49

subacute NOEL/NEL 19, 20, 28

subchronic or chronic NOAEL 6

skin or eye irritation 19, 28

skin or eye sensitization 19, 28


subchronic or chronic NOEL 25, 49

subchronic or chronic LOAEL 17

subchronic or chronic LD lo and TDlo 34

acute, subchronic, and chronic toxicity (rat) 47

General Ecological Effects (7 systems)

ecological disruption 10

ecosystem recovery potential 23

ecological effects benchmarks 32

AWQC 43

aquatic toxicity RQ 43

team rank chemical category 43

ecotoxicity 44

ecological effects RQ 45

effect on fertility (mammals, birds, plants, etc) 47

Systemic (Non-carcinogenic) or General Health Effects (30 systems)

acute LD50 and LC50 9, 29, 37

NOAEL and LOAEL 1, 43

LOEL 7

10-day health advisory (HA) 5

max. concentration level 8



18


number of "highly toxic" chemicals 16

potency, relative potency 18, 23

other toxic effects 35

type of effect 35, 36, 37, 38

unspecified 46

chronic MED 1, 2, 35

RfD/RfC 1, 5, 7, 9, 23, 33, 42, 43, 45, 48, 50

chronic toxicity severity rating (Rv) 2

systemic mammalian toxicity 10

irritation 11, 47, 51

sensitization 11, 47, 51

general acute or chronic toxicity 5, 8, 11, 33, 36, 44, 51

chronic ADI 27, 32

RQ (acute, chronic, cancer) 33, 41, 42, 43, 45

NAAQS 40

TPQ 42, 43, 45

neurotoxicity 29, 44

Toxic Substances Control Act chemical scoring system category (TSCA CSSC)

43

HWQC 43

chemical category human toxicity estimate (CCHTE)

43

human health structure activity rank (HHSATR) 43

Carcinogenicity/Mutagenicity/Genotoxicityc (36 systems)

carcinogenicity

weight, amount, or type of evidence; or probability

1, 2, 4, 6, 7, 9, 10, 17, 20, 22, 24, 25, 29, 33, 35, 39, 43, 44, 49

potency or slope factor (q1* or CPF) 1, 9, 42, 43, 50

ED10 1, 9, 10, 33, 35

qualitative potency or effects 2, 3, 4, 23, 25, 34, 48, 49



19

CERCLA hazard rank 7

carcinogen (yes or no) 11, 20, 47

URF (unit risk factor) 5, 7, 45

UCR 27

cancer potency RQ 41, 43

EEC guidance 28

unspecified endpoint 36, 51

number of carcinogens 16

mutagenicity

weight, amount, or type of evidence 1, 4, 17, 22, 51

severity or dose 1

effects 3, 4, 34

EEC guidance 28

bacterial or short-term tests 19, 20

mutagen (yes or no) 29, 47

genotoxicity

weight, amount, or type of evidence 1, 17, 24

severity or dose 1

genotoxin (yes or no) 11

Developmental/Reproductive Toxicity (19 systems)

unspecified 36, 44

weight, amount, type of evidence 1, 10, 22, 24, 51

severity, dose 1, 10

teratogenic effects 6, 34

effective dose 6, 17, 22, 25, 49

LOEL 7

developmental or reproductive toxin (yes, no) 11, 29, 39, 47

teratogen, EEC guidance 28

teratogenesis and fetotoxicity 38



20

Physical Hazard and Other Properties (15 systems)

ignitability, flash point 2, 11, 22, 28, 35, 37

boiling point 2, 11, 28, 35, 51

reactivity, explosivity 2, 11, 22, 28, 35, 37

corrosivity 22

odor/taste and appearance 22

pH 22

molecular weight 28, 48

melting point 28

relative density 28

vapor pressure 28, 48, 51

surface tension 28

water solubility 28, 48

fat solubility 28

oxidizing properties 28

ecological risk reduction potential (ERRP) 43

human health risk reduction potential (HHRRP)

43

radionuclide dose factors 15, 48

welfare effects (recreation, aesthetics etc.) 23

gross alpha 16

Sax toxicity rating 14, 16

presence on CAA Amendments list 42

low Kow 51

other dangers not covered by specific criteria

12

(a) refer to Table 1 for the corresponding systems(b) the number of systems listed in parentheses indicates how many of the evaluated systems consider someendpoint in the listed category(c) criteria could apply to ecological and/or human receptors, depending on the system. Often, this distinction is notclear in the system documentation.

21

The CERCLA HRS (EPA, 1990) evaluates human health effects based on acute toxicity and twocategories of chronic toxicity, which include cancer and non-cancer toxicological responses. The non-cancer score is based on the RfD, and the score for carcinogenicity is based on the q1

*. The higher ofthe two chronic toxicity scores is the score assigned for human toxicity. Acute toxicity data are used toscore human toxicity only in the absence of RfD and q1

* data.

The Ontario MOE Candidate Substances List for Bans or Phase-outs system (Socha et. al., 1992)includes mammalian acute lethality, sublethal effects on mammals and chronic effects such asteratogenicity, genotoxicity/mutagenicity and carcinogenicity for evaluating human health effects.

Not only is there a wide variation of endpoints selected for assigning a score or scores in the area ofhuman health effects, but the preference given to these endpoints may vary as well. For example, theHRS (EPA, 1990) prefers chronic toxicity data over acute data, whereas many other scoring systemsinclude both. Often, the purpose or application of the system affects the choice of criteria to beconsidered.

Potency and Severity of EffectsPotency. The dose required to elicit a toxic effect is referred to as potency. Data regarding the potencyof chemicals in test organisms are more frequently used to assess acute rather than chronic effects toterrestrial and aquatic plants and animals. For acute toxicity endpoints such as lethality, the doserequired to elicit the effect (i.e. potency) is commonly reported (e.g. LC50). Data regarding the potencyof chemicals required to elicit chronic effects such as mutagenicity or carcinogenicity, however, areoften limited.

Many ranking and scoring systems assign scores, particularly for carcinogenicity, based only on theweight of evidence that the chemical causes the effect in humans. Examples include:

• Michigan CMR (MDNR, 1987);• German Existing Chemicals of Environmental Relevance List I and II (BUA;

Behret, 1989a,b); and• The MOE Candidate Substance List for Bans or Phase-outs (Socha et. al., 1992).

It is helpful to distinguish between a highly potent carcinogen and a moderately potent one, but theavailability of potency data is limited. A few systems include both weight-of-evidence and potencydata. For example, GWU (Foran and Glenn, 1993) assigns a score according to a matrix whichincludes a weight-of-evidence classification as well as the potency factor (1/ED10) that has been used toassign reportable quantities for carcinogens by the CERCLA Section 102 program. In the EPA DfEUse Cluster Scoring System (EPA, 1993) weight-of-evidence is combined with the reportablequantity potency factor or the q1

*. The CERCLA HRS (EPA, 1990) assigns values based on weight-of-evidence and the q1

* or an estimate based on the ED10.

22

Severity. Severity is a function of both the type and magnitude of the effect. The type of effect (e.g. aspecific biochemical effect) reflects the mechanism of action as well as the target organ of a particularchemical. The magnitude of the effect (e.g. the percent change from normal of the biochemical effect)reflects the dose-response properties associated with a chemical (Environ, 1986).

Currently, there is no widely accepted method for incorporating severity into chemical ranking andscoring systems and many systems do not include measures of severity at all. This may yield misleadingresults. For example, two chemicals may have exactly the same NOEL of 0.3 mg/kg/day forneurotoxicity. Assigned scores based on the NOEL would be the same for these two chemicals, eventhough at similar doses one may cause a temporary narcotic effect and the other long-term irreversiblebrain damage. Below are examples of two systems that do include severity when scoring substances:

The Toxic Substances Control Act (TSCA) System (O'Bryan and Ross, 1988) includes severity inseveral parameters. For example, for nonlethal acute toxicity, the dose score is multiplied by theseverity score to arrive at a final score in this category. The severity of effects are scored according towhether the effect is life-threatening or severe, moderately serious, mild, or no effects are observed athigh doses.

The Michigan CMR (MDNR, 1987) also includes severity in several parameters. For example, ascore for "other toxicity" (chronic or subchronic) to terrestrial animals includes severity ranging fromadverse effects to severe effects. Examples of each effect classification are provided to guide decisionsin response to the question of how severe is the effect. For example, "moderate effects" include effectssuch as degenerative or necrotic changes with no apparent decrement of organ functions; or reversible,slight changes in organ function.

2.3.3 Criteria and Endpoints for Environmental Effects

Overview of Environmental EffectsEcological effects resulting from chemical exposure may occur in populations of organisms from manytrophic levels. These include both terrestrial mammalian and non-mammalian species, aquaticorganisms, and plants. Most ranking systems evaluate ecological effects through the use of toxicity testdata. Typically, surrogate species are selected for use in the evaluation of the potential environmentalhazards of chemicals.

When including toxicity information in a chemical ranking and scoring system, it is important to specifythe following information regarding the endpoints to be considered (EPA, 1989d):

• organism tested or observed;• nature of the effect;• concentration or dose needed to produce the effect;• duration of exposure needed to produce the effect; and

23

• environmental conditions under which the effects were observed.

Similar to tests used for evaluating human health effects, the durations for testing chemicals of ecologicalconcern include acute, sub-chronic and/or chronic exposure periods. Observations include lethality aswell as sub-lethal effects. Lethal doses directly disrupt important physiological functions and result indeath. Sub-lethal toxicity may have long-term physiological or behavioral impacts on a population.

Terrestrial effects. Terrestrial organisms commonly considered by chemical ranking and scoring systemsto be important receptors include mammals, non-mammals and plants. Most scoring systems do notmake a distinction between mammalian toxicity and human health effects. At least two of the chemicalranking and scoring systems examined use the same endpoint for human health effects as was used tocharacterize terrestrial mammalian toxicity (rodent acute toxicity). Terrestrial non-mammalian data areusually derived from tests on avian species such as the mallard duck or ring-neck pheasant. Data forterrestrial plants are sometimes included, but such data are limited. Other organisms may be includedas surrogates, but unfortunately, there is not a large data base of information regarding the effects ofchemicals on most terrestrial organisms.

Aquatic effects. The aquatic receptors of interest in most chemical ranking and scoring systems includenon-mammalian aquatic organisms and plants. The most common surrogates for evaluating the toxicpotential of chemicals include fish, Daphnia and algae. There is a fairly large data base of informationregarding the toxicity of chemicals to these organisms. The majority of available data are from short-term toxicity tests performed on these species, but in reality there is often greater concern regarding theeffects of long-term exposure and effects. Additionally, the types of organisms most often tested arenot necessarily those about which there is the most concern. Due to standardized test methods anddata availability, however, it is reasonable to include these commonly tested organisms in chemicalranking and scoring systems.

The endpoints most frequently included in ranking and scoring systems for aquatic toxicity are the LC50

and the EC50. The LC50 describes lethality and the EC50 describes the concentration at which there areobserved effects in 50% of the test population. The effects may be behavioral or physiological and mayindicate immobilization or changes in growth and reproduction, among others. Other endpoints, such asthe no-observable-adverse-effect-level (NOAEL) or regulatory standards such as ambient waterquality criteria (AWQC) may be included.

In the absence of data, there are over fifty high-quality quantitative structure-activity relationships(QSARs) for estimating biological effects in these three organisms (Clements, 1993). QSARs arediscussed further in section 2.2.5.

Endpoints used for Environmental Effects

24

Endpoints used for environmental effects for each system are presented in Appendix C and aresummarized in Table 2. (The use of criteria and endpoints is discussed further in section 3.2.1.) Thefollowing examples illustrate some of the types of data used to evaluate ecological effects:

The Schmallenberg System (Klein et. al., 1988; Weiss et. al., 1988) was developed by a group ofFrench and German officials and scientists and uses many endpoints to assess ecological effects,including acute, subacute and chronic effects in algae, plants, earthworms, birds and fish.

The GWU system (Foran and Glenn, 1993) includes acute and chronic toxicity to aquatic organisms aswell as terrestrial and avian species.

The EPA Toxics Release Inventory Environmental Indicators Methodology (Abt Assoc., 1992)scores ecological effects based on a matrix of aquatic toxicity and bioaccumulation. Aquatic toxicity isscored according to the LC50, life cycle/chronic NOAEL, or other measures such as the Acute orChronic Ambient Water Quality Criteria (AWQC). Bioconcentration is based on the water solubility,log Kow, or bioconcentration factor (BCF).

The EPA DfE Use Cluster Scoring System (EPA, 1993) scores ecological hazard based on theAWQC (acute and chronic) or Aquatic Toxicity Reportable Quantities.

The Michigan Critical Materials Register (MDNR, 1987) scores acute and chronic effects onaquatic organisms (fish, invertebrates, amphibians), acute and chronic effects on terrestrial animals, andplants.

The SIDS database (Auer, 1992) includes the following ecotoxicological data:

• acute toxicity to fish;• acute toxicity to daphnids;• acute toxicity to algae; and• terrestrial toxicity (if significant exposure in terrestrial environment or if aquatic toxicity

testing is not possible).

2.3.4 Measures of ExposureAs stated earlier, chemical ranking and scoring systems span a wide range of complexity. The simplestlevel is to use toxicity data alone. When exposure is considered, the systems become more complex. In fact, the level of complexity is primarily a function of the degree of sophistication used to estimateexposure. In a quantitative risk assessment, a dose to each receptor from each potential pathway isestimated on a site-specific basis using measured or modeled concentrations for each potentiallycontaminated environmental medium. This approach represents a high degree of sophistication and isnot used in most chemical ranking and scoring systems. In a simpler approach, production or usevolume or TRI emission data are often used as surrogates for dose. An intermediate level of

25

sophistication might employ a multimedia fate and transport model to estimate the distribution ofchemicals in the environment based on release data.

Many factors affect the potential for exposure to a chemical. These factors generally relate toproperties of the chemical, characteristics of the human or environmental receptors of concern, amountof the chemical available for exposure, and the behavior (fate, transport) of the chemical in theenvironment. Appendix C and Table 3 present the measures of exposure used by the ranking systemsevaluated. This is also discussed further in section 3.2.1. Examples of the use of these variousmeasures for exposure follow.

The WMS Scoring System (Könemann and Visser, 1988; Timmer et. al., 1988) assigns scores forenvironmental exposure according to use volume, percentage release to the environment, degradation inair, soil and/or water, relative occurrence in these media and bioconcentration. Exposure via productsis also scored, and this includes use patterns, exposure frequency and intensity of exposure.

The GWU system (Foran and Glenn, 1993) assesses exposure according to the bioaccumulation,persistence, and release or production volume.

The Michigan CMR (MDNR, 1987) scores chemical exposure according to bioaccumulation,persistence, and a few physicochemical properties such as flammability, reactivity, and corrosivity.

The Effluent Monitoring Priority Pollutants List (EMPPL) (Environment Ontario, 1987, 1988)includes environmental persistence, bioaccumulation and detection in the environment.

The German Beratergremium für Umweltrelevante Altstoffe (BUA) System (Behret, 1989a,b)includes bioaccumulation, persistence and production volume.

The EPA Toxics Release Inventory Environmental Indicators Methodology (Abt Assoc., 1992)uses facility-specific data and generic fate, transport, and exposure models to estimate a "surrogatedose", or the amount of chemical an individual might be exposed to. A separate evaluation isconducted for each release pathway allowing comparisons across media. The level of uncertainty isincluded in the scoring for exposure. Exposure of aquatic life is obtained by estimates of the ambientwater concentration.

26

Table 3. Endpoints Used for Scoring Exposure


Degradation or Transformation Potential (25 Systems)b

degradability in air, water, soil and/or sediment (t 1/2)

3, 4, 6, 9, 10, 20, 21, 22, 24, 25, 34, 48, 49

biodegradation 8, 19, 47

transformation 24, 45

oxidation 34

photolysis 9, 35, 47

hydrolysis 8, 9, 29, 34, 35

BOD 1/2 life, BOD5/COD, BOD28, BOD/ThOD 11, 12, 28, 29, 35

volatilization half-life 9

log Kow 9, 43

qualitative degree of persistence/ expert judgement

16, 43

source existence (yrs) 16

EPA persistence rating scale 14

Mobility/Partitioning (30 Systems)

adsorption (Kd, Koc) 8, 21, 34, 45

water solubility 1, 9, 11, 21

log Kow (P, Pow) 1, 3, 4, 6, 9, 11, 12, 19, 20, 21, 22, 24, 25, 28, 34, 43, 47, 49,51

BCF 1, 3, 4, 6, 9, 12, 17, 20, 22, 24, 25, 29, 39, 43, 49

BAF (some systems consider BCF=BAF) 10, 17, 22, 34

molecular weight (regarding bioconcentration) 11, 51

bioconcentration (yes/no) 45

vapor pressure 21

initial partitioning/transport 19, 24, 45, 47, 48, 51

evaporation 22

volatilization 9, 45

site-specific characteristics 16, 32

leaching potential 21

environmental transfer factors (e.g. soil to plants) 48



27

environmental spread/degree of mobility (qualitative)

28, 16

Estimated Dose, Environmental Occurrence, Concentration, or Releases (37 Systems)

Dose estimate

human exposure potential (mg/kg/day) 1

theoretical daily dose 2

estimated daily intake 32

chemical concentration, frequency or occurrence

exposure to contaminant or medium containing contaminant

2

exposure via migration pathway 9

projected levels in various media/compartments 19, 46, 47,

estimated ambient concentrations 1, 7, 23

frequency of occurrence/application 2, 14, 21, 23, 34

occurrence in air/soil/water 3, 4, 20, 27, 39

measured concentrations 15, 26, 32, 40, 50

concentration of chemical introduced 21

Use, production amounts, release or emission data

number of sites of discharge/use 34, 43

emission source data 36, 45

emissions estimates (rate,type,release height, temp.,flow rate,vent diameter)

36

TRI releases/transfers 5, 29, 42, 46

national emissions data (various sources) 7

release to environment (annual) 10, 24, 34, 41, 43, 44, 45

production volume 10, 11, 12, 19, 21, 24, 44, 51

import volume 11, 12, 51

waste volume 16

use volume 10, 20, 43

quantity on the market 28

release reduction potential 43



28

Exposure Frequency or Intensity (Receptor Characteristics) (11 Systems)

use pattern 19, 20, 51

population size/number potentially exposed 1, 23, 24, 27, 28, 32, 43

land use 32

distance from site 32

presence of critical environments 32

intensity of exposure 20, 23, 24

length of exposure 23

probability of exposure 24

frequency of exposure 20, 23

plurality of exposure 28

geographical extent of exposure (ecological) 23(a) refer to Table 1 for the corresponding systems(b) the number of systems listed in parentheses indicates how many of the evaluated systems consider someendpoint in the listed category

29

The SIDS database (Auer, 1992) includes the following data to assess sources and levels of exposure:

• production ranges (metric tons per annum);• categories and types of use;

and environmental fate and pathways:

• aerobic biodegradability;• abiotic degradability (hydrolysis and photodegradation by estimation); and• estimates of environmental fate, pathways and concentrations (including Henry's Law

constant, aerosolization, volatilization, soil adsorption and desorption).

2.3.5 Missing Data, Data Selection ApproachDue to the extensive time and resources required to perform thorough testing of each of the thousandsof chemicals in commerce, the available data necessary for complete assessment of the hazards of thesechemicals are limited. Therefore, chemical ranking and scoring systems must somehow work in theabsence of complete data sets for every chemical. An overview of the approaches for assigning valuesto criteria or endpoints used by the ranking systems evaluated can be found in Appendix C. Fourgeneral approaches were identified:

1. Assign one endpoint per criteria being assessed and estimate missing data (e.g. use oralLD50 for rats for acute human toxicity criteria). In the UT system (Davis et. al., 1993) there is only oneendpoint specified for each criterion, and missing data are estimated using QSARs, surrogatechemicals, or other structure-activity relationships (SARs).

2. Choose data from a hierarchy of endpoints, listed in order of preference, based on dataquality, appropriateness of test, etc. The Chemical Use Cluster Scoring Methodology (EPA,1993) uses a data hierarchy based on data quality for scoring the toxicity components. Preference isgiven to high quality data as specifically defined in the document. Data of medium or low quality maybe used in the absence of high quality data. When scoring non-carcinogens, the RfD and RQ areclassified as "high quality" data whereas chronic and sub-chronic NOAELs are "medium quality". Ifmultiple data are available within the high quality classification, then the data yielding the highest score(i.e. most conservative or health-protective) is used.

3. Choose the most conservative value from a pool of different endpoints. The GWU System(Foran and Glenn, 1993) is an example of a system where a group of endpoints could be consideredfor a given criterion, and the data yielding the most conservative (health-protective) result are chosen. If multiple data are available for a particular criterion (e.g. acute mammalian toxicity), then the dataresulting in the highest score are used. One drawback that should be noted for such a system is that itdoes not encourage further testing of compounds, because it is unlikely that a score for a chemical canbe lowered by filling data gaps.

30

4. Assign cutoff (or trigger) values to a large number of criteria and select a chemical if one ora specified number of the criteria are met. Several systems designed to select chemicals forregulatory action or for further study do not provide an overall score for a chemical. Rather, a chemicalis selected if it meets certain criteria. Examples of this type of system include the Michigan CMR(MDNR, 1987), where a chemical is selected for regulatory action if it receives a certain score; and theCandidate Substance List for Bans or Phase-Outs (Socha et. al., 1992), where certain cut-off valuesare used to determine if a substance is to be considered as "persistent", "bioaccumulative" and/or"toxic". The primary list consists of substances considered to be all three of these.

For approaches where required data are not available, it is necessary to somehow estimate the missingdata. SARs, QSARs, default values, and/or ad hoc expert judgement have been used in the absence ofexperimental data. SARs and QSARs are often used in many chemical scoring and ranking systems toestimate chemical properties for which data are lacking. SARs qualitatively relate known data from achemical with similar molecular or molecular fragment structure, and therefore similar expectedbiological activity or other properties, to the chemical with missing data. QSARs are obtained primarilythrough regression analysis which results in mathematical equations representing correlations. Chemicalstructure or physicochemical properties, such as molecular weight and octanol-water partitioningcoefficients, are typically compared with chemical toxicity or fate (Hermans, 1989). The limitations ofthese methods must be weighed against the impact of missing data on the results of a chemical rankingor scoring system.

Examples of systems that use SARs QSARs for predicting the effects of chemical exposure arediscussed below.

The Schmallenberg system (Klein et. al., 1988; Weiss et. al., 1988) defines the minimum data setrequired for priority-setting. Data on mutagenesis and acute aquatic toxicity are essential and may beestimated by QSARs when necessary. Other important missing data can be estimated from QSARs,but any information, even if it is semi-quantitative in nature is considered.

The German (BUA) scoring system (Behret, 1989a,b) assigns negative scores where data areunavailable. These scores are based largely on SARs. The use of negative scores occurred for only afew chemicals evaluated by this system.

The UT system (Davis et. al., 1993) requires a quantitative assessment for many of the endpointsincluded in the algorithm. This is obtainable through the use of expert judgement, SARs or QSARs forestimation.

The EPA DfE Use Cluster System (EPA, 1993) allows a score for human health hazard potential tobe assigned by a structure activity team. Ecological hazard potential may be scored according toQSAR concentrations of concern.

31

Another approach to filling data gaps results in the assignment of default values in the absence of dataor reliable QSARs or SARs. The European Communities (EC) system (van de Zandt and vanLeeuwen, 1992) uses default values and flags the data to identify when default values are used.

A few systems use expert judgement to give qualitative or semi-quantitative scores to chemicals (e.g.high, med., low). An example of a system using this approach is the EPA Unfinished Business Report(Morgenstern et. al., 1987). The project was described as being based on "informed judgement". Quantitative information available within the EPA was collected on the various environmental problemsaddressed, but because the qualitative judgement was extensive, the study results were not consideredto be scientifically "reproducible".

2.3.6 Aggregation and WeighingThe majority of the chemical ranking and scoring systems reviewed include human health andenvironmental effects as well as exposure parameters. The manner in which these parameters arecombined to give an overall ranking or risk categorization varies tremendously and significantly effectsthe final result. Scores may be added, multiplied, divided or not combined at all. Scores may beweighted differently for human versus environmental effects (e.g. Screening Methodology forPollution Prevention Targeting, EPA, date unknown), acute versus chronic effects (e.g. EPA, 1990)or exposure via water versus exposure via air (e.g. Klein et. al., 1988; Weiss et. al., 1988). Examplesof various aggregation and weighing methods follow.

Chemical selection in the Michigan CMR (MDNR, 1987) is based on combinations of scores in thevarious parameters. For example, selection for the Register may occur if a chemical scores a "5" in twoor more criteria. These criteria include all of the health, environmental and exposure parameters wherenone are considered more important than the others.

The Nordic system (Gjøs et. al., 1989) selects chemicals with a score of "high" in an effects category(health or environmental) regardless of the exposure score or chemicals with an effect score of"medium" and an exposure score of "high".

The WMS Scoring System (Könemann and Visser, 1988; Timmer et. al., 1988) calculates ten scoresfor ten different combinations of exposure and effects. Examples of such combinations include 1)Exposure via air and general toxicity for mammals, carcinogenicity, and mutagenicity; and 2) Exposurevia products and general toxicity for mammals, carcinogenicity, and mutagenicity.

In the MOE Candidate Substance List for Bans and Phase-outs (Socha et. al., 1992) substances areselected on the basis of combinations of scores in toxicity, persistence and bioaccumulation. Forexample, a substance is placed on the primary list if it scores a "10" in any one of the seven toxicitycategories and has a persistence of a half-life of 50 days and it has a bioconcentration factor of morethan 500.

32

In the ATSDR Priority List (ATSDR, 1992) scores are assigned for each of three criteria: frequencyof detection at National Priorities List (NPL) sites, toxicity, and potential for human exposure. Thetotal score is simply the sum of the scores assigned for each of these parameters.

Once scores are obtained for each of the various exposure and effects parameters, chemicals may beplaced into categories, such as selected or non-selected chemicals, or they may be ranked in order ofpriority.

Categorization may be accomplished by determining an overall concern score of high, medium or low,or by selecting chemicals according to combinations of numerical scores in the exposure and effectscategories. For example:

The Schmallenberg system (Klein et. al., 1988; Weiss et. al., 1988) scores chemicals and places themin one of three categories: immediate action, observation, or no action.

The GWU Method (Foran and Glenn, 1993) assigns scores of high, medium or low in several toxicitycategories, release and production, persistence, and bioaccumulation. A chemical is selected as acandidate for Sunsetting if 1) it scores "high" in any toxicity category and "high" in release andproduction (excluding pesticides) or 2) it scores "high" in any acute or chronic toxicity category and"high" in persistence or bioaccumulation (including pesticides).

The WMS Scoring System (Könemann and Visser, 1988; Timmer et. al., 1988) categorizes chemicalsinto three areas of priority by plotting an image point on a two-dimensional diagram. The image point isdetermined after each chemical has been assigned a rank for exposure and effects.

The MOE Candidate Substances List for Bans or Phase-outs (Socha et. al., 1992) assigns numericalscores for environmental transport, persistence and bioaccumulation as well as seven categories fortoxicity. Specific combinations of scores in these categories result in the inclusion of the chemical on theprimary or secondary list for appropriate action.

The scoring systems which are designed to rank chemicals often include an algorithm for combiningindividual scores for toxicity and exposure in order to determine an overall score which represents arelative indicator of hazard for each chemical. Theseranking systems then prioritize chemicals in order of their potential hazard to human and environmentalhealth. For example:

The UT Method (Davis et. al., 1993) uses an algorithm which includes both additive and multiplicativeparameters to determine an overall hazard value for each chemical. The chemical rank indicates itshazard relative to the other chemicals that are scored.

33

The ATSDR Method (ATSDR, 1992) places 275 substances in order of hazard potential based on thetotal score. The total score is obtained by the following formula:

Total score = NPL frequency + Toxicity + Potential for human exposure

The EPA DfE Use Cluster Scoring System (EPA, 1993) arrives at an overall chemical score asfollows:

Chemical score = Human risk reduction potential+Ecological risk reduction potential+EPA interest level

The average chemical score for the entire cluster is then added to the cluster score for pollutionprevention potential to give an overall cluster score which forms the basis for prioritizing clusters.

Whether a method results in categorization or ranking of chemical substances will depend on thespecific goals of the system. Systems developed for regulatory objectives frequently categorizechemicals by selecting those that pose the greatest potential threat for regulation and/or monitoring. Systems for prioritization for other purposes may rank chemicals in order to focus research or otheractivities on those chemicals highest on the list.

2.3.7 Other IssuesOther issues that can be considered in chemical ranking and scoring include:

• indirect ecological impacts, such as greenhouse gases, acid rain, eutrophication, andstratospheric ozone depletion;

• effect of chemical mixtures;• application to life cycle assessment, specifically, life cycle impact assessment; and• social or economic impacts.

34

Indirect Ecological ImpactsThe release of chemicals into the environment can have direct effects on human health and theenvironment, but indirect effects may result as well. Certain substances are known to affect abioticcomponents of the environment and ultimately result in large scale ecological disruption. For example, afew substances (e.g. carbon dioxide) are considered "greenhouse gases" because they are believed tohave the potential to change the global climate. Other examples of chemicals that may cause theseindirect impacts are sulfur and nitrous oxides, which contribute to acid rain, chlorofluorocarbons(CFCs), which contribute to stratospheric ozone depletion, and nitrous oxides and volatile organiccompounds (VOCs) which contribute to tropospheric (low atmosphere) ozone formation. Anotherexample is eutrophication of surface waters caused by excess nutrient inputs.

Most chemical ranking and scoring systems do not include these types of indirect impacts as scoringcriteria. The hazards posed by substances causing these effects have not been quantified in suchmanner that they can be easily included. This is not to say that the risks are not significant or that theyshould be ignored in chemical scoring. Ayres (1993), in an analysis of the relationship between naturalresources and economic growth, concludes by asserting that "those activities that are most likely tointerfere with natural climatological and nutrient cycling processes" should be of primary importance toboth economic development and environmental protection policy.

Foran and Glenn (1993) list several reasons why there is no mechanism currently available forpredicting or quantifying large scale ecological disruption. First, due in part to a lack of understandingof how hazardous compounds affect the structure and function of ecosystems it is difficult to predict theadverse impacts at an ecosystem or global level. Additionally, hazardous substances may interact withcomponents of ecosystems, but due to a lack of understanding of the fate and transport characteristicsof these substances, their effects on ecosystems are difficult to predict. It is important to haveinformation about the use and release patterns of a chemical in order to predict large-scale ecologicaldisruption, but this information is not generally available when a chemical is created and tends to changeover time.

The GWU system (Foran and Glenn, 1993) considers evidence for ecological disruption whenselecting candidates for Sunsetting. The data available are mostly derived from studies of effects innatural ecosystems, but other chemical-specific studies may be considered as well. Laboratory studiesor theoretical evaluations that suggest ecological disruption potential must be supported by predicteduse and release patterns of the substance.

The TRI Environmental Indicators Methodology (Abt Assoc., 1992) has not yet included scoringcriteria for these effects, but it notes a major EPA project aimed at determining the risks associatedwith CFCs and their alternatives. The model is complicated but could perhaps be used to evaluate therisks posed by the emissions of these substances. Methods are also suggested for developing anindicator for tropospheric ozone. This would involve identifying all VOCs reported in the TRI that areconsidered to be ozone precursors and to sum the releases of these compounds from each facility.

35

Alternatively, policy changes which would require reporting of VOCs and nitrous oxides (NOx) as agroup would help in the development of an indicator. The developers of the system are also examiningmodels designed to determine the exposure resulting from particle deposition to evaluate the risksposed by airborne chemicals which are deposited into other media, such as surface water.

Because chemical ranking and scoring systems have focused on direct toxicity to humans and otherorganisms in the environment, indirect, large-scale impacts such as those discussed above have rarelybeen taken into account. Impact assessment approaches have been developed, and the challenge willbe to incorporate these with the direct toxicological impacts. The field of life cycle assessment,discussed below in section 2.3.7.3 is one place where this challenge is being addressed.

Effect of Chemical MixturesThe majority of toxicity tests are conducted for single chemicals, usually at high doses. Organisms inthe environment, however, are more typically exposed to low levels of several chemicals rather than toone chemical at a time. Chemicals may interact with one another to elicit effects that are different fromthe combined effects of the individual chemicals. The effects may be additive, where the total effectsapproximately equal the sum of the individual chemical effects; synergistic, where the total effects aregreater than the additive effects; or antagonistic, where the total effects are less than the additive effects. Other types of chemical interaction can also occur that result in unanticipated effects. There are limiteddata regarding the effects of chemical mixtures. One data source developed by the EPA (MIXTOX)contains summary information from and literature citations on studies of toxicological interactions,primarily for binary mixtures (EPA, 1992).

The chemical ranking and scoring systems evaluated do not consider the effects of chemical mixtures. Because they are not usually site-specific, they are designed to evaluate the general potential harm fromsubstances on a chemical by chemical basis without regard to the effects of mixtures. It is difficult topredict the effect of chemical mixtures on humans and environment. If incorporation of mixture toxicityinto chemical ranking and scoring systems is a goal, however, there is some evidence to suggest thatreal world situations involving many organic chemicals are likely to result in approximate additivetoxicity (McCarty and MacKay, 1993).

Life Cycle AssessmentRecently, much attention has been focused on the methodology of LCA, a holistic approach toevaluating the human health and environmental burdens associated with a product or process life cycle. LCAs are tools for evaluating the effects on the environment associated with products, processes oractivities (SETAC, 1993). A full LCA includes a quantitative inventory of resource and energy inputsand pollutant outputs and some form of impact assessment. A life cycle impact assessment is a processfor assessing the potential and actual effects of environmental loadings identified in the inventory(SETAC, 1993). Chemical ranking and scoring could become an essential element in the developmentof tools for assessing the impacts to health and the environment from chemical releases throughout thelife cycle of products.

36

LCA has several uses, although its uses are more limited if some form of impact assessment is notincluded. LCA can be used for internal product improvements, for designing new products, for settingpublic policy on products and materials, and for environmental labeling. Clearly, the different uses ofLCA create different needs for impact assessment. An LCA used for internal product improvement,for example, might simply use the inventory component and operate on a "less-is-best" approach. AnLCA used for setting public policy on materials or products would need to include some framework forassessing and comparing the significance of environmental releases and resource and energy use of thedifferent products and materials being compared.

One of the systems evaluated was developed for life cycle impact assessment. The EPS-Enviro-Accounting Method (Steen and Ryding, 1992) was developed to assess the health and ecologicaleffects associated with the entire life-cycle of a product, process, or activity. The main objective of theEPS Method is to provide one overall economic measure of resource depletion and potential healthand environmental impacts throughout a life cycle.

In this system, values are assigned to impacts on the environment in terms of five 'safe-guard subjects'(human health, biodiversity, production, resources & aesthetic values) according to willingness to pay torestore them to normal status. Emissions, use of resources, and other human activities are then valuedaccording to their estimated contribution to the changes in these safeguard subjects. The information onenvironmental impacts originates from LCA-based inventory of the materials/process under study.

Economic ConsiderationsChemical ranking and scoring systems are usually designed to be fairly quick and simple, limiting thetime and resources required for assessment to a minimum. It is usually beyond the scope of thesesystems to determine potential economic impacts caused by releases of a particular chemical.

One system that does include economic considerations is the EPS Enviro-Accounting system (Steenand Ryding, 1992) which is intended for life cycle impact assessment. Impacts on five safeguardsubjects are valued on a relative scale which is based on the willingness to pay for avoiding undesiredeffects. Environmental load units (ELUs) which are standard monetary values, are assigned accordinglyto score impacts. As another example, the Unfinished Business Report (Morgenstern et. al., 1987)considers monetary estimates of damage when categorizing environmental problems as they relate towelfare effects.

37

Social ValuesEconomic impacts can be included under the broader category of social welfare or social values. SETAC (1993, p. 20) outlines a variety of "social welfare impact categories"; some of these issues orimpacts possibly relevant to chemical ranking and scoring include:

• demographic impacts, including fertility and mortality, morbidity, migration;• sociopolitical impacts, including legal, governmental;• social impacts, including quality of life;• community impacts, including land use, physical appearance of community, community

satisfaction;• sociocultural impacts, including social justice, aesthetics, environmental values; and• psychosocial impacts, including anxiety or stress.

Only a few systems include social considerations, per se, in the scoring or selection criteria. Oneexample, The Michigan Critical Materials Register (MDNR, 1987) includes aesthetic considerationssuch as taste, odor and appearance as criteria for scoring chemicals within one parameter. TheUnfinished Business Report (Morgenstern et. al., 1987), although not specifically a chemical rankingsystem, ranks issues related to social welfare such as recreation, losses in aesthetics and non-uservalues. The EPS Enviro-Accounting method (Steen and Ryding, 1992) includes aesthetic values asone of the five safeguard subjects for which impacts are evaluated.

38

SECTION 3: DISCUSSION

3.1 RISK ASSESSMENT PRINCIPLES APPLIED TO CHEMICAL RANKING ANDSCORING

Chemical ranking and scoring systems are typically intended to be fairly simple and quick methods fordetermining the health and environmental hazards posed by the use and release of chemical substances. Although not intended to provide a quantitative assessment of risk, the majority of the systems revieweddo rely on the basic principles of risk assessment for chemical ranking and scoring.

Chemical risk is a product of both toxicity and exposure. Most chemical ranking and scoring systemsinclude measures of both toxicity and exposure and, in this way, are similar to quantitative riskassessment methods. The major difference is the extent to which the exposure assessment isperformed.

Exposure assessment in a quantitative risk assessment involves an analysis of contaminant releases,identification of exposed populations, identification of all potential exposure pathways, and estimation ofexposure point concentrations and contaminant intakes. The exposure assessment results in an estimateof the magnitude, frequency and duration of actual or potential human exposures through variouspathways expressed as a total dose. None of the chemical ranking and scoring systems reviewedinclude a detailed site-specific quantitative risk assessment, although the (TRI) EnvironmentalIndicators Methodology (Abt Assoc., 1992) approaches this level of detail.

The system developed by Jones et. al. (1988) is one of the few systems that does not include anyexposure assessment. The stated purpose of the system, however, is to estimate only the relativetoxicological potency for hazardous substances rather than overall hazard. Another example is themethod developed by the EPA (1993b) for demonstrating the relative hazard of emissions undersection 112(g) of the Clean Air Act. As required, offsetting emission decreases must be considered"more hazardous" than emission increases. For categorizing emissions, toxicity criteria alone areconsidered. The Sax Toxicity Ratings (Sax and Lewis, 1989) is also based only on chemical propertiesposing a physical hazard and toxicological characteristics. The other systems reviewed include somemeasure of exposure, although the endpoints used to characterize exposure vary significantly.

The final step in a baseline risk assessment is risk characterization. Here, chemical toxicity data arecombined with potential exposure levels for the receptors of interest at a site to arrive at a quantitativeestimate of the risk that receptors will suffer adverse effects. Such site-specific characterization is notgenerally performed in chemical ranking and scoring, but most of the systems reviewed do combinetoxicity and exposure in some manner to score, select, or prioritize chemicals. As discussed in Section2.3.6, the manner in which toxicity and exposure are combined for an overall assessment also variessignificantly.

39

3.2 SIMILARITIES AND DIFFERENCES AMONG SYSTEMS REVIEWED

3.2.1 Most Common Effects and Exposure Endpoints Used The effects and exposure categories specified by the systems were difficult to compare. They weredivided, as best could be done, into categories of commonly recognized criteria for presentation inTables 2 and 4 and in Appendix C. More information is available in Appendix A , where the criteria,subcriteria, and endpoints as defined by each system are outlined. Some criteria and endpoints werecommonly used in many systems while others were unique to a particular system. The most commonlylisted endpoints for major classes of effects criteria include:

• carcinogenicity, mutagenicity, genotoxicity, most often characterized by the weight,type, or amount of evidence that a chemical would elicit that effect;

• systemic (non-carcinogenic) or general health effects, most commonly characterizedby chronic or subchronic Rfd or RfC values;

• aquatic toxicity, most often quantified by acute LC50 and EC50 data;• mammalian toxicity, most often quantified by acute LD50 and LC50 data;• developmental/reproductive toxicity, again most often measured by the weight, type,

or amount of evidence;• physical hazard, most often characterized by ignitability, boiling point, and reactivity;• plant toxicity, most commonly measured by EC50 data;• terrestrial non-mammalian toxicity, most often characterized by acute LD50; and• general ecological effects, with no specific endpoint used by more than one system.

The most commonly listed endpoints for major classes of exposure criteria include:

• degradation or transformation potential: most commonly measured by half-life in theenvironment and some type of BOD data;

• mobility and partitioning, most often characterized by Kow and BCF;• estimated dose, environmental occurrence, concentration, or amount released,

most commonly measured by annual releases to environment and production volume;and

• exposure frequency or intensity, relates to potential receptors and usually is measuredby population size or number potentially exposed.

The classes of criteria are listed in both cases in decreasing order from the most to the least oftenincluded as components of the evaluated systems.

3.2.2 Ranking versus CategorizationAs mentioned in Section 2.3.6, there are two major types of chemical ranking and scoring systems;those that rank chemicals and those that categorize chemicals. Examples of categories that may beused include substances of high, medium or low concern or selected chemicals, non-selected chemicals

40

and chemicals for further review. Ranking systems, however, usually derive overall scores and rankchemicals relative to one another based on these scores.

In developing a consensus framework for chemical scoring, a decision must be made whether to use thesystem to rank chemicals, select or categorize chemicals, or rank and select chemicals. To a certainextent, this decision depends on the purpose of the system, but one disadvantage for a system that onlycategorizes or selects chemicals is that those selected substances cannot be further prioritized. Forexample, suppose such a system were used to choose a subset of chemicals that are considered to beof high concern from a regulatory perspective. If 200 chemicals meet the criteria to be considered highconcern, there may need to be a means of further prioritizing these substances before regulatory policydecisions could be made. van de Zandt and van Leeuwen (1992) suggest that a ranking system ispreferable to classification methods for this reason.

Three of the systems reviewed assign chemical scores which are not used to rank or categorize thechemicals. For example, the TRI Environmental Indicators Methodology (Abt Assoc., 1992) scoresTRI chemicals to reflect the impacts of the releases and transfers. These numbers can then becompared from one year to the next to monitor changes in impacts. Chemicals are not ranked orselected.

Several systems assign chemical scores and rank the substances according to relative hazard. Thepriority list of substances prepared by the ATSDR (1992) is the result of ranking and scoring, in orderof priority, chemicals commonly found at NPL sites.

Chemical ranking may be used for selecting chemicals for further review, regulatory action or someother activity. An example is the Ranking System for the Clean Water Act (Poston and Prohammer,1985; Cornaby et. al., 1986) in which chemicals were ranked and then selected as candidates forpossible inclusion on or deletion from the Priority Pollutant List. There are seven systems reviewedwhich result in both ranking and selection of chemicals. Thirteen of the systems evaluated resulted in the selection or categorization of chemicals, mostly forregulatory purposes (ten systems), without first ranking them. An example of such a system is thatdeveloped by ICF (1990) for the EPA which selects chemicals as "High Priority Pollutants" if certainexposure and effects criteria are met. Another example is the Michigan CMR (MDNR, 1987) which isderived from scoring chemicals for various exposure and effects criteria. Chemicals are then selectedfor inclusion in the Register if they receive a score a "5" (the maximum score) in two or more criteria orscore an additive level of "15" or greater for all criteria. Chemicals are chosen for more detailedscrutiny if other similar criteria are met.

3.2.3 Quantitative versus Qualitative EndpointsChemical ranking and scoring systems are basically quantitative in nature, oftentimes resulting in rankingor selection based on numeric cut-off values for various criteria. The assignment of scores for specificendpoints, however, is not always based on quantitative data. Many of the systems reviewed include a

41

combination of both qualitative and quantitative endpoints. Particularly for chronic health effects, suchas carcinogenicity, qualitative data are essential. A large number of the systems which includecarcinogenicity as a scoring criterion utilize weight-of-evidence and/or strength-of-evidence information. Sometimes, as in the GWU system (Foran and Glenn, 1993), these qualitative data are combined withquantitative potency estimates where available. Qualitative data are also used frequently in the absenceof quantitative data for certain criteria. For example, the BUA system (Behret, 1989a,b) assignsnegative scores where qualitative information is utilized in the absence of a desired quantitativeendpoint. When there is no data such as the Kow or bioconcentration factor (BCF) regardingbioaccumulation, then a score of -1 is assigned if there is no suspicion of bioaccumulation or a -2 ifbioaccumulation potential is suspected. Qualitative data, such as SARs, are frequently used to fill datagaps.

3.2.4 Assigning Scores to Endpoint DataFor those systems where there is agreement on which particular endpoint to use, there still exists avariety of approaches to assign corresponding numerical scores. To illustrate this, BCF is used as anexample of an endpoint included in several systems and scored in a variety of ways. Table 4 presentsthe numerical scores assigned to BCF or log BCF by eight systems. For comparison purposes, allBCF data are converted to log BCF and the scores are normalized on a scale of 0 to 1 correspondingto the minimum and maximum value assigned by the system. This information is also presentedgraphically in Figure 1.

A few items are worthy of note. First, there is a noticeable lack of agreement; the eight systemsrepresented here use seven different schemes to assign values to BCF data. Second, the majority ofsystems use some kind of step function to assign values to endpoint data (i.e. if log BCF is between 3and 4, then assign a value of x). One exception is the UT system, which uses a continuous linearfunction with upper and lower bounds. The systems that use step functions use different numbers ofsteps. The BUA system has only two; it simply assigns a minimum score if the log BCF is less than 2and a maximum score if it is greater than 2. Several others use five steps. Finally, it can be seen thatdifferent systems assign varying levels of significance to BCF values. The BUA system considers a logBCF of 2.1 to be of maximum significance, while, for this same BCF value, the Michigan CMR assignsa relatively low value and the UT system assigns a mid-range score.

Similar evaluations could be useful for all endpoints included in a chemical ranking and scoring system,but are beyond the scope of this report.

42

TABLE 4. Bioconcentration Factor (BCF) Scores for Several Systems

System name(System no.)

BCF Log BCF Score Normalizedscore a

Candidate Substance List for Bans orPhase-outs (6);Effluent Monitoring Priority PollutantList (EMPPL) (25);Canadian AcceleratedReduction/Elimination of Toxics(ARET) Scoring Protocol (49)

>15,000500 - 15,000

20 - 500<20

>4.22.7 - 4.21.3 - 2.7

<1.3

10 7 4 0

1.00.70.4

0

Michigan Critical Materials Register(MCMR) (22)

>100,00010,000 - 99,999

1,000 - 9,999100 - 999

<1 - 99

>54 - 53 - 42 - 3

<2

S(sufficient)5310

1.0 b

0.710.430.14

0

WMS Scoring System (20) >31.5 - 3

<1.5

210

1.00.5

0

Chemical Scoring System for Hazardand Exposure Identification (TSCA)(24)

>1000200 - 1,000

100 - 20010 - 100

<10

>32.3 - 32 - 2.3

1 - 2<1

97530

1.00.780.560.33

0

TRI Environmental IndicatorsMethodology (1)

>10,000 1,000 - 10,000

100 - 1,00010 - 100

1 - 10<1

>4 3 - 42 - 31 - 20 - 1

<0

50,0005,000

500505

0.5

1.0.79.57.36.15

0

Existing Chemicals of EnvironmentalRelevance (BUA) I & II (3,4)

>100<100

>2<2

20

1.00

UT Method (29) >41 - 4

<1

2.50.5logBCF+0.5

1

1.0(calc)

0.4(a) normalized on a scale from 0 to 1 by dividing each score by the maximum score for that system(b) normalized assuming a maximum score of 7

43

Figure 1

44

3.3 GENERAL STRENGTHS AND WEAKNESSES OF EVALUATED SYSTEMS

A detailed critique of all fifty-one systems reviewed is beyond the scope of this report. Instead, ageneral discussion is presented for consideration either in choosing an existing system, or in developinga new or hybrid system. Van de Zandt and van Leeuwen (1992) suggest that a priority setting systemshould be:

• quick to use as a screening tool to identify priority substances for further review;• systematic and computerized;• transparent, with methodologies described in detail for clarity to the user• flexible;• accurate, avoiding too many false positives or negatives and working with available

data to give consistent results;• based on a scientifically justifiable framework and generally accepted methodologies;

and• based on exposure and effects, with emphasis on long-term effects.

These aspects, along with a few additional issues we have identified, are discussed below:

Ease of use, complexity. The EPA chemicals of potential concern selection process is simple andstraight forward when concentration data are available. The required toxicity data, however, arereadily available for only a limited number of chemicals. The TRI indicators methodology is verycomplex and amounts to practically conducting a quantitative risk assessment at every facility thatreports TRI releases. The time and resources required is an important deciding factor in choosing ordeveloping a method. Also, the mathematical or mechanical processes of scoring and ranking shouldbe straight forward enough so that the system is transparent to the user. It should easy to understandthe results given a basic knowledge of the chemicals and data used.

Flexibility. A system should have enough flexibility so the user can augment or simplify it, add orsubtract components, add new data as they become available, or change the weighing for variouscriteria while the system maintains its integrity. The UT system is an example of a relatively flexiblesystem.

Availability of data. Many systems require data that either are not readily available or frequently requireestimation methods. As much of the required data as possible should be peer-reviewed. The UseCluster Scoring System (EPA, 1993) draws on a number of possible endpoints, arranged in ahierarchy for high, medium and low quality data. This data selection approach allows the user to drawfrom a much wider range of data, and should result in fewer instances of missing data.

Comparability to other systems. Ranking and scoring results, especially when regulations are involved,should be comparable across regulatory programs and across industries. Ideally, they should also be

45

comparable on an international basis. One example where international comparability could beimportant is where trade issues are involved. Three of the evaluated systems (the Canadian SubstanceList for Bans or Phase-outs, the EMPPL system, and the ARETS Scoring Protocol) use the samescoring criteria, based on the MOE scoring criteria, although their selection processes differ slightly. Otherwise, there is little consistency among the evaluated systems.

Number of chemicals the existing system has been demonstrated on. Some systems may seem like agood idea, but may disintegrate under the pressure of a large number of chemicals. A few of theevaluated systems were presented only in theory, with no demonstration using real chemical data. Others have been demonstrated or used on hundreds of chemicals. For example, over 700 chemicalswere ranked by the CERCLA 104 Priority List system (ATSDR, 1992), and van de Zandt and vanLeeuwen (1992) addressed 2000 HPV chemicals in the Priority Setting of Existing ChemicalSubstances system. Applicability to various classes of chemicals. Ions, metals and radionuclides pose problems that arevery different from organic compounds. For example, many QSAR or SAR estimation methods do notapply to inorganic chemicals. The CERCLA HRS system (Hallstedt et. al., 1986) was intended onlyfor ranking organic compounds, but has been modified to include radionuclides in the site rankingprocess (Hawley & Napier, 1985).

Reproducibility or subjectiveness of scoring methods. Results from a panel of experts or whenqualitative estimations are used may not be reproducible at different times or when applied by differentusers. For example, the WMS Scoring System relies an a panel of experts to score chemicals; if usedin another application, different experts could arrive at different scores for the same chemicals. Inaddition, a system should be based on scientific principles and an understanding of risk assessmentmethodologies.

Completeness. All potential and important effects should be included in a system. The followingcomponents should be included, or the decision not to include a component should be made clear:

• the purpose and application of the ranking and scoring system;• the human health criteria and endpoints included;• the criteria and endpoints included for environmental effects;• whether measures of exposure are included;• the data selection approach and handling of missing data;• the use of aggregation and weighing of different health and environmental impacts;• methods of accounting for chemical potency and severity of effects; and• inclusion of other impacts or issues.

46

SECTION 4: CONCLUSIONS

This review of various chemical ranking and scoring systems demonstrates that there is currently noconsensus regarding an appropriate framework for evaluating adverse impacts to human health and theenvironment from exposure to chemical substances. There is much variation in the degrees ofsophistication, the types and numbers of endpoints incorporated, the data selection approach and themanner of weighing and/or combining scores to arrive at a final assessment.

Clearly, the purpose for which the ranking or scoring tool is intended will affect these factors. Althoughthe intended purpose of a ranking or scoring system can influence its design, it would be useful todevelop a standard framework for chemical ranking and scoring that is flexible enough to be adaptedfor most purposes. A standardized system for ranking and scoring chemicals is desirable for severalreasons. It could assist business and regulatory agencies to target their pollution prevention efforts;enable environmental policy and business decisions to be made on similar, objective, and scientifically-based methods; provide consistency nationally across regulatory programs as well as internationally;make use of and build on previous experience gained from the development of existing systems; allownew information to be incorporated while maintaining system integrity; and save on time and expenseincurred by continuing to develop new systems from scratch with every new application.

Due to the large variation among ranking systems in terms of criteria, endpoints, data selectionapproaches, and scoring methods, one logical way to approach a consensus on methods would requirethe following steps (illustrated in Figure 2):

• Develop a consensus on the overall framework for chemical ranking and scoring,including its component parts: criteria, endpoints, data selection approach, and scoringformula;

• Work toward agreement on the component parts, such as:

- which criteria should be included in any chemical ranking and scoring system- which endpoint(s) should be used to measure or score that criteria- what data selection approach should be used to select or estimate data for the

criteria- how should criteria scores be weighted and combined to reach an overall score

or rank for each specific chemical (if an overall score per chemical is the goal)

47

Figure 2

48

- what level of sophistication in exposure estimation is appropriate - from totalamounts produced, used or released, to multimedia environmental fate models,to site-specific models including estimates of dose over time

• Reassemble these elements into the agreed-upon framework, providing flexibility to usesome or all of the components for specific purposes.

The process of developing the consensus framework should involve all of the significant stakeholders -government agencies, chemical manufacturers and users, environmental and consumer groups, andacademic researchers. This will create greater acceptance of the results.

Given the current interest in chemical ranking and scoring, and the many potential applications, an effortto develop a consensus framework for chemical ranking and scoring would be an importantcontribution to reducing chemical risk.

49

REFERENCES

Abt Associates, Inc. (1991). Ranking the Relative Hazards of Industrial Discharges to POTWsand Surface Waters. Prepared for U.S. Environmental Protection Agency, Office of PolicyAnalysis.

Abt Associates, Inc. (1992). Toxics Release Inventory Environmental Indicators Methodology. (DraftReport). By Abt Associates, for U.S. Environmental Protection Agency, Office of PollutionPrevention and Toxics, Washington, DC.

Agency for Toxic Substances and Disease Registry (ATSDR) (1992). Support Document: TheCERCLA 104 Priority List of Hazardous Substances That Will Be The Subject ofToxicological Profiles. U.S. Public Health Service, Department of Health and HumanServices, Washington DC.

ARET Criteria Sub-Committee (1993). ARET Criteria Sub-Committee Report, revision dateSeptember 27, 1993.

Ashby, J. et. al. (1990). A Scheme for Classifying Carcinogens. Regulatory Toxicology andPharmacology, 12, 270-295.

Auer, C.M. (1992). Memorandum, "The OECD "SIDS" Program", June 17, 1992 (updated fromOECD document S6/CK/BIAC91.181/4.9.91) Director, Existing Chemicals AssessmentDivision. U.S. Environmental Protection Agency, Office of Pesticides and Toxic Substances,Washington, DC.

Ayres, R.U. (1993). Cowboys, Cornucopians and Long-Run Sustainability. Ecological Economics, 8,189-207.

Beckvar, N. & L. Harris (1992). Coastal Hazardous Waste Site Review, September, 1992.Hazardous Materials Response and Assessment Division, NOAA, ORCA, Seattle, WA.

Behret, H. (Ed.) (1989a). Existing Chemicals of Environmental Relevance. GDCh-AdvisoryCommittee on Existing Chemicals of Environmental Relevance. VCH, New York.

Behret, H. (Ed.) (1989b). Existing Chemicals of Environmental Relevance II, Selection Criteriaand Second Priority List. GDCh-Advisory Committee on Existing Chemicals ofEnvironmental Relevance. VCH, New York.

Bouchard, D. (1991). Review of Region VII TRI Strategy. (Memo, EPA Region VII).

50

Bureau of National Affairs, Inc. (BNA) (1991) Criteria for Identifying High Risk Pollutants,Environment Reporter, Washington, DC, pp. 463-465.

Canadian Labor Congress (CLC) (1992). A Critique of the Ontario Hazard Assessment System.CLC Environment Bureau, Ottawa, Ontario.

Clements, R.G, J.V. Nabholtz, D.W. Johnson & M. Zeeman (1993). The Use and Application ofQSARs in the Office of Toxic Substances for Ecological Hazard Assessment of NewChemicals. In Environmental Toxicology and Risk Assessment, ASTM STP 1179.

Cornaby, B.W. et. al. (1986). Results of Implementation of a Chemical Ranking System. Preparedby Battelle, for U.S. Environmental Protection Agency, Washington, DC.

Crutcher, M.R., & F.L. Parker (1990). A Classification System for Hazardous Chemical Wastes.Superfund 90, Hazardous Materials Control Research Institutes, 11th Annual NationalConference, 222-225.

Davis, G.A. et. al (1993). Chemical Ranking for Potential Health and Environmental Impacts. University of Tennessee, Center for Clean Products and Clean Technologies, Knoxville, TN.

Droppo, J.G., Jr. et. al. (1989). Multimedia Environmental Pollutant Assessment SystemApplication Guidance, Volume 2 - Guidelines for Evaluating MEPAS Input Parameters.Battelle Memorial Institute, Pacific Northwest Laboratory, Richland, WA.

Environ Corporation (1986). Examination of the Severity of Toxic Effects and Recommendationof a Systematic Approach to Rank Adverse Effects. Prepared for U.S. EnvironmentalProtection Agency, Office of Environmental Criteria and Assessment, Cincinnati, OH.

Environment Ontario (1987). The Effluent Monitoring Priority Pollutants List. Ontario Ministry ofthe Environment, Hazardous Contaminants Coordination Branch. ISBN 0-7729-2784-7.

Environment Ontario (1988). The Effluent Monitoring Priority Pollutants List, 1988 Update.Ontario Ministry of the Environment, Hazardous Contaminants Coordination Branch. ISBN 0-7729-5450-X.

Environmental Monitoring and Services, Inc. (EMS) (1985). Technical Background Document toSupport Rule Making Pursuant to CERCLA Section 102 (Volumes 1-2). Prepared forU.S. Environmental Protection Agency.

51

Foran, J.A. & B.S. Glenn (1993). Criteria to Identify Chemical Candidates for Sunsetting in theGreat Lakes Basin. The George Washington University, Environmental Health and PolicyProgram, Department of Health Care Sciences, Washington, DC.

Gjøs, N., M. Møller, G.S. Hægh, & K. Kolset (1989). Existing Chemicals: Systematic DataCollection and Handling for Priority Setting. Center for Industrial Research, Oslo, Norway.Nordic Council of Ministers, Copenhagen.

Gustafsson, L. & E. Ljung (1990). Substances and Preparations Dangerous for the Environment:A System for Classification, Labelling and Safety Data Sheets. Final Report from a NordicWorking Group, Nordic Council of Ministers, Copenhagen.

Halfon, E., & M.G. Reggiani (1986). Notes on Ranking Chemicals for Environmental Hazard.Environ. Sci. Technol. 20, 1173-1179.

Hallstedt, P.A., M.A. Puskar, & S.P. Levine (1986). Application of the Hazard Ranking System to thePrioritization of Organic Compounds Identified at Hazardous Waste Remedial Action Sites.Hazardous Waste and Hazardous Materials, 3(2), 221-232.

Hawley, K.A., & B.A. Napier (1985). A Ranking System for Hazardous Sites & With MixedRadioactive and Hazardous Wastes, Proceedings of the Fifth DOE Environmental ProtectionInformation Meeting. U.S. Department of Energy, Office of Operational Safety, PacificNorthwest Laboratory, Richland, WA.

Hermens, J.L.M. (1989). Quantitative Structure-Activity Relationships of Environmental Pollutants. InHandbook of Environmental Chemistry. Hutzinger, O. (Ed.), Volume 2E, Springer Verlag,Berlin, pp. 111-162,.

Hushon, J.M. and M.R. Kornreich (1984). Scoring Systems for Hazard Assessment, in: HazardAssessment of Chemicals: Current Developments, Volume 3, J. Saxena (Ed.). AcademicPress, Inc., Orlando.

Hutchinson, W.R., & J.L. Hoffman (1983). A Ground Water Pollution Priority System (N.J.Geographical Open File, Report No. 83-4). Division of Water Resources, Trenton, NJ.

ICF Incorporated (1990). Targeting Pollution Prevention Opportunities Using the 1988 ToxicsRelease Inventory. Prepared for U.S. Environmental Protection Agency, Office of Policy,Planning and Evaluation, Pollution Prevention.

ICF Incorporated (1993). Summary and Comparison of Five Chemical Scoring Systems.Submitted to: Chemical Manufacturers Association, Washington, D.C.

52

International Joint Commission's (IJC) Binational Objective Development Committee (1989). TheGreat Lakes Water Quality Agreement Standard Methods and Annex 1, Lists ofSubstances.

Jones, T.D., P.J. Walsh, A.P. Watson, B.A. Owen, L.W. Barnthouse, & D.A. Sanders (1988).Chemical Scoring by a Rapid Screen of Hazard (RASH) Method. Risk Analysis, 8(1), 99-118.

Kincaid, L.E. & J.E. Bartmess (1993) Evaluation of TRI Releases in Indiana, Louisiana, Ohio,Tennessee and Texas. University of Tennessee, Center for Clean Products and CleanTechnologies, Knoxville, Tennessee.

Klaassen, C.D. & D.L. Eaton (1993). Principles of Toxicology. In Casarett and Doull's Toxicology:The Basic Science of Poisons, Fourth Edition. M. Amdur, J. Doull, C.D. Klaassen (Eds.).McGraw Hill, Inc., New York.

Klein, W., W. Kördel, A.W. Klein, D. Kuhnen-Clausen, & M. Weiss (1988). Systematic Approachfor Environmental Hazard Ranking of New Chemicals. Chemosphere, 17, 1445-1462.

Könemann, H. & R. Visser (1988). Selection of Chemicals With High Hazard Potential: Part 1:WMS-Scoring System. Chemosphere, 17, 1905-1919.

Landis, W.G., J.S. Hughes & M. Lewis (Eds.). American Society for Testing and Materials,Philadelphia, pp. 56-64.

Laskowski, P.A., C.A.I. Goring, P.J. McCall, & R.L. Swann (1982). Principles of Environmental RiskAnalysis: Terrestrial Environment. in Environmental Risk Analysis for Chemicals, R. Conway(Ed.), pp. 198-240. Van Nostrand Reinhold, New York.

McCarty, L.S. & D. MacKay (1993). Enhancing Ecolotoxicological Modeling and Assessment.Environmental Science and Technology, 27(9), 1719-1728.

Michigan Department of Natural Resources (MDNR) (1987). Critical Materials Register. (Criteriaand Support Documents).

Morgenstern, R., D. Clay, G. Emison, R. Hanmer, & M. Williams (1987). USEPA UnfinishedBusiness: A Comparative Assessment of Environmental Problems, Volume 1. U.S.Environmental Protection Agency, Washington, DC.

O'Bryan, T.R. & R.H. Ross (1988). Chemical Scoring System for Hazard and Exposure Identification.J. Toxicol. Env. Health, 1, 119-34.

53

Organization for Economic Co-operation and Development (OECD) (1986). Existing Chemicals:Systematic Investigation, Priority Setting and Chemicals Reviews. Paris, France.

Poston, T.M. & L.A. Prohammer (1985). A Ranking System for Clean Water Act Section 307(a)List of Priority Pollutants. Prepared by Battelle for U.S. Environmental Protection Agency,Washington, DC.

Radian Corporation (1990). The Source Category Ranking System: Development andMethodology. Prepared for U.S. Environmental Protection Agency, Office of Air QualityPlanning Standards, Chemicals and Petroleum Branch, Research Triangle Park, NC.

Rechard, R.P., G.G. Wilkinson, & J.D. Schreiber (1991). User's Manual for SRS88: Site RankingSystem for Chemical and Radioactive Waste (SAND87-2815, UC-721). Sandia NationalLaboratories, Albuquerque, NM.

Rechard, R.P., M.S.Y. Chu, & S.L. Brown (1988). SRS: Site Ranking System for HazardousChemical and Radioactive Waste. Sandia National Laboratories, Albuquerque, NM.

Sampaolo, A. & R. Binetti (1986). Elaboration of a Practical Method for Priority Selections and RiskAssessment among Existing Chemicals. Reg. Toxicol. & Pharmacol., 6, 129-154.

Sampaolo, A. & R. Binetti (1989). Improvement of a Practical Method for Priority Selections and RiskAssessments Among Existing Chemicals. Reg. Toxicol. & Pharmacol., 10, 185-195.

Sax, N.I. & R.J. Lewis (1989). Dangerous Properties of Industrial Materials. Van Nostrand Reinhold, New York.

Scala, R.A. (1993). Risk Assessment. In Casarett and Doull's Toxicology: The Basic Science of Poisons, Fourth Edition. M. Amdur, J. Doull, C.D. Klaassen (Eds.). McGraw Hill, Inc.,New York.

Silka, L.R., & T.L. Swearingen (1978). A Manual for Evaluating Contamination Potential ofSurface Impoundments. (EPA 570/9-78-003). U.S. Environmental Protection Agency, Officeof Drinking Water, Washington, DC.

Society of Environmental Toxicology and Chemistry (SETAC) (1993). A Conceptual Framework forLife-Cycle Impact Assessment, J. Fava, et. al., Eds., A Workshop Report, February 1-7,1992, Sandestin, Florida. Sponsored by SETAC and the SETAC Foundation forEnvironmental Education, Inc.

54

Socha, A.C., T. Dickie, & R. Aucoin (1992). Candidate Substances List for Bans or Phase-Outs. Ontario Ministry of the Environment.

Steen, B., & S.O. Ryding (1992). The EPS Enviro-Accounting Method. Swedish EnvironmentalResearch Institute (IVL), Göteborg, Sweden.

Strenge, D.C., S.R. Peterson, & S. Sager (1989). Chemical ta Base for the MultimediaEnvironmental Pollutant Assessment System (MEPAS): Version 1. U.S. Department ofEnergy, Battelle Memorial Institute, Pacific Northwest Laboratory, Richland, WA.

Timmer, M., H. Könemann, & R. Visser (1988). Selection of Chemicals With High Hazard Potential:Part 2: WMS-Scoring System. Chemosphere, 17, 1921-1934.

U.S. Department of Defense (DOD)(1992). User's Manual for the Defense Priority Model (FY 93version, Interim Draft). Prepared by Earth Technology Corporation and ERM ProgramManagement Company, for U.S. Department of Defense, Office of Deputy Assistant Secretaryof Defense (Environment), Washington, DC.

U.S. Environmental Protection Agency (EPA)(1978). Measuring Air Quality: The New PollutantsStandards Index. U.S. Environmental Protection Agency, Office of Policy Analysis,Washington, DC.

U.S. Environmental Protection Agency (EPA)(1986). Screening Procedure for Chemicals ofImportance to the Office of Water. U.S. Environmental Protection Agency, Office of Healthand Environmental Assessment, Washington, DC.

U.S. Environmental Protection Agency (EPA)(1989a). Toxic Chemical Release Inventory RiskScreening Guide (Vol. 1). U.S. Environmental Protection Agency, Office of Toxic Substances,Washington, DC, EPA 560/2-89-002.

U.S. Environmental Protection Agency (EPA)(1989b). Risk Assessment Guidance for Superfund,Volume I, Human Health Evaluation Manual. U.S. Environmental Protection Agency, Officeof Emergency and Remedial Response. EPA/540/1-89/002.

U.S. Environmental Protection Agency (EPA)(1989c). Technical Background Document to SupportRule Making Pursuant to CERCLA Section 102 (Volume 3) U.S. Environmental ProtectionAgency, Office of Solid Waste and Emergency Response.

55

U.S. Environmental Protection Agency (EPA)(1989d). Risk Assessment Guidance for Superfund,Volume II, Environmental Evaluation Manual, Interim final. U.S. Environmental ProtectionAgency, Office of Emergency and Remedial Response. (EPA/540/1-89/001).

U.S. Environmental Protection Agency (EPA) (Dec. 14, 1990). Hazard Ranking System; Final Rule.Federal Register 55 (241): 51532-51667 (40CFR Part 300).

U.S. Environmental Protection Agency (EPA) (1992). MIXTOX: An Information System onToxicologic Interactions for the MS-DOS Personal Computer, Version 1.5, User's Guide.U.S. Environmental Protection Agency, Environmental Criteria and Assessment Office,Cincinnati, OH.

U.S. Environmental Protection Agency (EPA) (1993a). Chemical Use Clusters ScoringMethodology (Draft). U.S. Environmental Protection Agency, Office of Pollution Preventionand Toxics, Chemical Engineering Branch, Washington DC.

U.S. Environmental Protection Agency (EPA) (1993b). Hazardous Air Pollutants: ProposedRegulations Governing Constructed, Reconstructed and Modified Major Sources (40CFR Part 63).

U.S. Environmental Protection Agency (EPA) (date unknown). Screening Methodology for PollutionPrevention Targeting. U.S. Environmental Protection Agency, Office of Toxic Substances.

U.S. Environmental Protection Agency (EPA) (date unknown). TSCA's TRI Chemical RiskAssessment Pre-Screening Methodology. U.S. Environmental Protection Agency, Office ofToxic Substances, Existing Chemical Assessment Division.

Waters, R.D., M.R. Crutcher & F.L Parker (1993). Hazard Ranking Systems for Chemical Wastesand Chemical Waste Sites. In Hazard Assessment of Chemicals, Volume 8. J. Saxena (Ed.).Taylor and Francis, Washington, D.C.

Weiss, M., W. Kördel, D. Kuhnen-Clausen, A.W. Lange, & W. Klein (1988). Priority Setting ofExisting Chemicals. Chemosphere, 17, 1419-1443.

Whelan, G. et. al. (1987). The Remedial Action Priority System (RAPS): MathematicalFormulations. U.S. Department of Energy, Battelle Memorial Institute, Pacific NorthwestLaboratory, Richland, WA.

Whelan, G. et. al. (1992). Overview of the Multimedia Environmental Pollutant Assessment System(MEPAS). Hazardous Waste & Hazardous Materials, 9 (2), 191-208.

56

van de Zandt, P.T.J., & C.J. van Leeuwen (1992). A Proposal for Priority Setting of ExistingChemical Substances. Netherlands Ministry of Housing, Physical Planning and theEnvironment.